See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/324654165

Nonlinear Consequences of Promotive and Prohibitive Voice for Managers'

Responses: The Roles of Voice Frequency and LMX

Article  in  Journal of Applied Psychology · March 2018

DOI: 10.1037/apl0000326

CITATIONS READS

0 379

4 authors:

Xu Huang Erica Xu
Hong Kong Baptist University Hong Kong Baptist University
77 PUBLICATIONS   2,502 CITATIONS    12 PUBLICATIONS   115 CITATIONS   

SEE PROFILE SEE PROFILE

Lei Huang Wu Liu

Auburn University The Hong Kong Polytechnic University
16 PUBLICATIONS   60 CITATIONS    29 PUBLICATIONS   447 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Workplace Incivility View project

Leader humor and employee outcomes View project

All content following this page was uploaded by Wu Liu on 17 September 2018.

The user has requested enhancement of the downloaded file.

Journal of Applied Psychology
Nonlinear Consequences of Promotive and Prohibitive
Voice for Managers’ Responses: The Roles of Voice
Frequency and LMX
Xu Huang, Erica Xu, Lei Huang, and Wu Liu
Online First Publication, June 25, 2018. http://dx.doi.org/10.1037/apl0000326

CITATION
Huang, X., Xu, E., Huang, L., & Liu, W. (2018, June 25). Nonlinear Consequences of Promotive and
Prohibitive Voice for Managers’ Responses: The Roles of Voice Frequency and LMX. Journal of
Applied Psychology. Advance online publication. http://dx.doi.org/10.1037/apl0000326
 Journal of Applied Psychology
© 2018 American Psychological Association 2018, Vol. 1, No. 999, 000
0021-9010/18/$12.00 http://dx.doi.org/10.1037/apl0000326

Nonlinear Consequences of Promotive and Prohibitive Voice for Managers’

Responses: The Roles of Voice Frequency and LMX

Xu Huang and Erica Xu Lei Huang

Hong Kong Baptist University Auburn University

Wu Liu
The Hong Kong Polytechnic University
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Departing from past research on managers’ responses to employee voice, we propose and examine a
nonlinear linkage between promotive/prohibitive voice and managers’ evaluations of voicers (i.e.,
manager-rated voicers’ promotability and overall performance). Drawing from social persuasion theory,
we theorize that managers tend to give more positive evaluations to employees who engage in a moderate
frequency of promotive/prohibitive voice than those who either rarely speak up or speak up very
frequently. In Study 1, based on a sample from a Chinese bank, we found that leader-member exchange
quality (LMX) moderated the inverted U-shaped linkage of prohibitive voice with manager-rated
promotability of voicers, whereas the frequency of promotive voice was not related to promotability,
irrespective of levels of LMX. In Study 2, using employee-reported voice frequency, rather than the
manager-rated measures adopted in Study 1, we largely replicated the main findings of Study 1 based on
a sample from an information technology firm in the United States. In Study 3, using another U.S.
sample, from a financial services firm, we found that manager-perceived voice constructiveness mediated
the curvilinear interactive effect of prohibitive voice (rather than promotive voice) and LMX on
managers’ evaluations of employees’ overall performance.

Keywords: promotive voice, prohibitive voice, LMX, consequences of voice, social persuasion

Challenging voice, defined as verbal expressions of opinions,
ideas, or concerns to constructively change and modify current
operations (Burris, 2012; Morrison, 2014), is believed to improve
organizational or unit functioning (Howard-Grenville, 2007; Li,
Liao, Tangirala, & Firth, 2017; Van Dyne & LePine, 1998). It
remains undetermined, however, whether employees engaging in
such voice behavior tend to receive favorable or unfavorable
responses from managers. A recent meta-analysis of 24 empirical
studies suggested that managers’ responses depend on the voice
content (Chamberlin, Newton, & LePine, 2017). Specifically, al-
though promotive voice (i.e., voice focused on suggesting new
solutions; Liang, Farh, & Farh, 2012) leads to managers’ more
Xu Huang and Erica Xu, Department of Management, Hong Kong
Baptist University; Lei Huang, Department of Management, Auburn Uni-
versity; Wu Liu, Department of Management and Marketing, The Hong
Kong Polytechnic University.
Xu Huang and Erica Xu contributed equally to this work, and the order
of authorship was determined alphabetically. This work has been supported
by the Research Grants Councils of the Hong Kong Special Administrative
Region, People’s Republic of China, (Project 592913) to Xu Huang. We
thank Jason Shaw and Subrahmaniam Tangirala for their helpful comments
on earlier versions of this article.
Correspondence concerning this article should be addressed to Xu
Huang, Department of Management, Hong Kong Baptist University, Kow-
loon Tong, Hong Kong. E-mail: xuhuang@hkbu.edu.hk
positive appraisals of voicers’ overall performance, prohibitive
voice (i.e., voice focused on pointing out current problems; Liang
et al., 2012) leads to managers’ more negative evaluations. Al-
though these meta-analytic findings are enlightening, recent voice
research has suggested that managers’ responses to employee
voice are determined not only by the content of voice (e.g.,
messages with or without a solution), but also by the characteris-
tics of voicers (e.g., expertise of voicers) and the voicing context
(e.g., organizational norms; Whiting, Maynes, Podsakoff, & Pod-
sakoff, 2012). These factors, however, were not taken into account
in Chamberlin et al.’s (2017) meta-analysis.
To provide a more complete explanation of managers’ responses
to challenging voice, our study investigates when and why pro-
motive/prohibitive voice is positively associated with managers’
evaluations of voicers, which refers to as managers’ appraisals of
employees’ potential value for the organization and overall per-
formance (Burris, 2012, p. 852). We draw from social persuasion
theory (e.g., Briñol & Petty, 2009; Jones, Sinclair, & Courneya,
2003; Petty & Cacioppo, 1986) to suggest that, in organizational
settings, the effects of promotive voice and prohibitive voice on
managers’ evaluations of voicers may not follow a simple linear
pattern, as depicted in Chamberlin et al.’s (2017) study. Rather, we
propose employees’ frequency of voicing (an observable voicer
characteristic) and their leader-member exchange quality with
their supervisors (LMX, a relational context) as two salient vari-
ables that regulate the effects of promotive voice and prohibitive
voice on managers’ evaluations.

1
 2 HUANG, XU, HUANG, AND LIU
Specifically, frequency of voice, defined as how often an em-
ployee takes the initiative to raise various work-related issues to
his or her immediate supervisor, ranging from never doing it to
voicing very frequently, signals a salient characteristic of voicers:
the extent to which they make efforts to initiate changes that may
benefit the organization. We propose that frequency of promotive
and prohibitive voice has an inverted U-shaped relationship, rather
than a linear relationship, with managers’ evaluations. This is
because managers tend to appreciate and reward employees who
exhibit a moderate level of challenging voice more than those who
show either few such initiatives or overly persistent efforts to
challenge the status quo (e.g., Fuller, Marler, Hester, & Otondo,
2015; Grant, Parker, & Collins, 2009; Lam, Rees, Levesque, &
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Ornstein, in press; Milliken, Morrison, & Hewlin, 2003; Schilit &
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Locke, 1982). Building on recent works on the effects of voice
content (Chamberlin et al., 2017; Maynes & Podsakoff, 2014;
Whiting et al., 2012), we further suggest a refined prediction of the
differential nonlinear impacts of promotive voice and prohibitive
voice on managers’ responses. By proposing a nonlinear effect of
voice frequency on managers’ evaluations, we offer a new way to
address the mixed findings regarding the effects of challenging
voice on managers’ evaluations (Burris, 2012; Chamberlin et al.,
2017; Seibert, Kraimer, & Crant, 2001; Van Dyne & LePine, 1998;
Whiting, Podsakoff, & Pierce, 2008).
Moreover, following social persuasion theory (Briñol & Petty,
2009; Menon & Blount, 2003), we cast LMX as a salient relational
context that exerts an overriding influence on managers’ responses
to promotive/prohibitive voice. We theorize that whereas manag-
ers in high LMX contexts tend to give positive evaluations to
voicing employees irrespective of the types of voice and voice
frequency levels due to ingroup favoritism, there is an inverted
U-shaped relationship between the frequency of promotive/prohib-
itive voice and managers’ evaluations in low LMX contexts. Our
model thus challenges the idea that challenging voice, in general,
and prohibitive voice, in particular, results in managers’ less
favorable evaluations, especially for outgroup employees (Menon
& Blount, 2003; Whiting et al., 2012).
Finally, the cognition in persuasion model, an extension of
social persuasion theory (Albarracín, 2002; Albarracín & Wyer,
2001; Hart et al., 2009), posits that cognitive processes, such as
message recipients’ cognitive assessments of the beneficial out-
comes of messages, play a central role in explaining how various
social persuasion factors (e.g., source, content, and context char-
acteristics) jointly influence receivers’ responses to persuasion.
LMX
Frequency of
Manager-
promotive voice
perceived voice
constructiveness
Frequency of
prohibitive voice
Figure 1. Proposed theoretical model. LMX ⫽ leader-member exchange quality.
Drawing from this theory, we propose and examine the mediating
role of manager-perceived voice constructiveness, defined as the
extent to which managers regard the voiced issues as making
positive contributions to the organization (Whiting et al., 2012). In
so doing, we contribute to the voice literature by theorizing and
demonstrating the critical role of cognitive mechanisms explaining
the consequences of challenging voice in organizational settings.
In short, we advance the literature of the consequences of voice
by examining an integrated model that shows what (voice content),
when (voice frequency and LMX), and why (voice constructive-
ness) challenging voice influences managers’ evaluations of voic-
ers by testing our model in two culturally different nations (China
and the United States) and by including both manager-reported
(Studies 1 and 3) and employee-reported (Study 2) voice behavior
in our analyses. Figure 1 illustrates our proposed research model.
Theory and Hypotheses
Social Persuasion and Consequences of Challenging
Voice
Social persuasion theory suggests that the content of messages
(e.g., how the messages are framed) influences the reactions of
message recipients (McGuire, 1985). Receivers are more receptive
to messages with convincing evidence (Isenberg, 1986; McCros-
key, 1969), clearly stated solutions (Hovland & Mandell, 1952),
and positive framing (e.g., Levin, Schneider, & Gaeth, 1998). In
line with this reasoning, Chamberlin et al.’s (2017) meta-analytic
study demonstrated that when the content of challenging voice is
framed as expressing opportunities to enhance organizational func-
tioning by doing new things in new ways in the future (i.e.,
promotive voice; Liang et al., 2012), managers tend to appraise
voicers’ overall performance more favorably. By contrast, when
the content of challenging voice is framed as pointing out prob-
lems with current practices and inhibiting wrongdoings to improve
organizational functioning (i.e., prohibitive voice; Liang et al.,
2012), managers tend to give less favorable performance evalua-
tions to the voicing employees. Although both promotive and
prohibitive voice are motivated by the same desire for well-
intended change and are driven by strong commitment to and
identification with the organization, the differences in their con-
tents may result in managers’ differential judgments and assess-
ments of the value and potential beneficial outcomes of voice,
Managers’ evaluations of
voicers
Studies 1 and 2
• Manager-rated
promotability
Study 3:
• Manager-rated overall
performance evaluation
 FREQUENCY OF VOICE AND MANAGERS’ RESPONSES
which in turn influence their evaluations of voicers’ overall per-
formance (Chamberlin et al., 2017).
Accumulated evidence in the social persuasion research has
suggested, however, that message recipients’ assessments of mes-
sages’ beneficial outcomes and their subsequent reactions are
affected not only by the content, but also by source and context
characteristics (McGuire, 1985). In a seminal work on social
persuasion, for example, Petty, Cacioppo, and Goldman (1981)
showed that the content of the message (e.g., strength of argu-
ments) gave way to the source characteristic (e.g., expert status) in
influencing message recipients’ judgments of the quality of the
message within the context of low thinking, suggesting that pe-
ripheral cues can play an important role in shaping social persua-
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
sion processes. Extending this idea to the voice context, Whiting et
This document is copyrighted by the American Psychological Association or one of its allied publishers.
al. (2012) proposed specific conditions that influence managers’
evaluations of the voice and the voicers. Using a series of labora-
tory studies, Whiting et al. (2012) demonstrated, for example, that
source, message, and context characteristics independently influ-
ence managers’ assessments of the voiced issues and thus their
evaluations of voicers’ overall performance.
Because of the nature of laboratory studies, Whiting et al.’s
(2012) model is particularly relevant to managers’ reactions to a
single “incident” of voice. In reality, however, challenging voice is
not an one-off endeavor. Employees may voice out challenging
issues and interact with their supervisors over a relatively long
period of time. Hence, managers’ evaluations of voicing employ-
ees are shaped not only by how employees frame their voice
(promotive vs. prohibitive), but also by employees’ habit of voice
behaviors (Lam et al., in press) and by employees’ ongoing social
interactions with their supervisors (Dutton & Ashford, 1993). We
therefore propose that, in organizational settings, the effects of
promotive and prohibitive voice on managers’ evaluations may not
follow a linear pattern, as depicted in Chamberlin et al.’s (2017)
study but instead are influenced by voicers’ frequency of voicing
that reflects important characteristics of voicers and their LMX
quality with managers, which captures a salient voicing context at
work.
Frequency and Content of Challenging Voice
In the workplace, some employees may remain silent; some may
speak up occasionally; others may frequently voice up issues that
challenge the status quo (Morrison & Milliken, 2003). Prior re-
search has suggested that individual characteristics, such as work
experience, conscientiousness, and proactive personality, predict
how often employees speak up at work (Li, Liang, & Crant, 2010;
Thomas, Whitman, & Viswesvaran, 2010). In a recent paper, Lam
et al. (in press) developed a theoretical framework to depict how
employees may form stable “habits” of voicing. They argued that
some individuals may lack situational awareness and controllabil-
ity and are concerned about efficiency. As a result, these individ-
uals tend to develop a “habit” of voicing and engage in voice
behaviors more frequently than others. According to social per-
suasion theory, certain characteristics of message senders can
communicate important cues that trigger message recipients’ judg-
mental shortcuts or heuristics of processing the messages, deter-
mine message recipients’ favorable or unfavorable thinking about
the messages, or serve as arguments for or against the persuasive
attempts (Briñol & Petty, 2009; Petty & Cacioppo, 1986). Thus, in
3
line with this theory, we regard frequency of voice as an easy-to-
observe characteristic of voicers that may provide salient periph-
eral cues, shaping managers’ assessments of promotive and pro-
hibitive voice. We anticipate that low, moderate, and high
frequency levels may signal qualitatively different characteristics
of the voicer, causing an inverted U-shaped relationship between
the frequency of promotive and prohibitive voice and managers’
evaluations of the voicing employee.
Compared to employees exhibiting moderate frequencies of
promotive/prohibitive voice, a low frequency may signal to man-
agers that the voicing employees are exerting relatively little effort
to initiate changes and challenge the status quo to improve the
organization’s functioning. This informative characteristic of voic-
ing employees may provide salient cues that exert a negative
influence on managers’ assessments of voice from these employ-
ees, regardless whether the voice is framed as promotive or pro-
hibitive (Eagly, Wood, & Chaiken, 1978). This is because man-
agers tend to degrade employees who do not take the initiative to
engage in their work and contribute to the organization (Fuller et
al., 2015; Grant et al., 2009). Managers may thus form less
favorable assessments of the voiced issues from employees with
low voice frequency and may be less likely to reward their voice
behavior with positive evaluations.
A moderate frequency of promotive/prohibitive voice may in-
form managers that the voicing employees are exerting relatively
more effort to initiate changes and challenge the status quo for the
organization than those who rarely voice up. This informative
characteristic of the voicers may lead managers to make more
positive judgments of their voiced issues and reward them with
more favorable evaluations, regardless whether the voice is pro-
motive or prohibitive. This is because managers in general tend to
appreciate and reward employees who take the initiative to con-
tribute to the organization (Burris, Detert, & Romney, 2013; Fuller
et al., 2015; Grant et al., 2009; Whiting et al., 2008; Whiting et al.,
2012).
High frequency of promotive/prohibitive voice, in contrast, may
result in managers’ unfavorable responses. Compared to employ-
ees who engage in a moderate frequency of promotive/prohibitive
voice, a high voice frequency may signal that the voicing employ-
ees are being relatively more forceful in their efforts to challenge
the status quo. Instead of predicting a linear and stronger positive
effect of promotive/prohibitive voice on managers’ evaluations,
we suggest that managers may not view promotive/prohibitive
voice positively at high frequencies. Rather, this informative char-
acteristic of voicing employees might cause managers to form
unfavorable judgments of voiced issues from these employees.
This is because managers tend to resist employees who persistently
and frequently challenge the status quo, and thus they may devalue
their voice (e.g., Milliken et al., 2003; Schilit & Locke, 1982).
Also, as Lam et al. (in press, p. 12) have theorized, employees who
habitually engage in challenging voice tend to “be viewed as
‘loose cannons’ . . . who ‘shoot from the hip’,” causing managers
to discredit their views. This is because managers may think that
these employees have not given sufficient thought to the voiced
issues, and thus they may see the employees’ voice as less valuable
and constructive. Hence, high frequency of voice provides a salient
cue that might exert a negative influence on managers’ assess-
ments of challenging voice from these employees and thus their
evaluations of the voicers, irrespective of the voiced content.
 4 HUANG, XU, HUANG, AND LIU
Taken together, we posit that voice frequency plays a key role in
determining managers’ responses. Specifically, compared to low
and high frequencies of promotive/prohibitive voice, we predict
that moderate frequencies are more likely to drive managers to
reward voicers by giving them favorable evaluations.
Hypothesis 1a: There is an inverted U-shaped relationship
between the frequency of promotive voice and managers’
evaluations of voicing employees.
Hypothesis 1b: There is an inverted U-shaped relationship
between the frequency of prohibitive voice and managers’
evaluations of voicing employees.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Although voice frequency may serve as a salient source char-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
acteristic that shapes managers’ responses to challenging voice,
Chamberlin et al.’s (2017) meta-analytic study suggested that
different contents of challenging voice lead managers to evaluate
voicers differently. Building on the insights from Chamberlin et
al.’s (2017) study, but departing from the linear perspective of
previous studies, we argue that the curvilinear effect of the fre-
quency of promotive voice on managers’ evaluations may follow
a different pattern from that of prohibitive voice. We posit that
both types of voice may be positively associated with managers’
evaluations at low-to-moderate frequency levels, as managers tend
to appreciate and reward employees for making efforts to chal-
lenge the status quo to improve organization functioning (Van
Dyne & LePine, 1998). At high frequency levels, although we
expect that both promotive voice and prohibitive voice tend to
have a negative impact on managers’ evaluations, we predict that
the negative effect of prohibitive voice is stronger than that of
promotive voice. This is because the future-oriented nature and
constructive tone of promotive voice tend to reinforce managers’
appreciation of employees’ efforts to initiate changes for the
organization (Chamberlin et al., 2017; Liang et al., 2012), miti-
gating managers’ negative responses to such voice, even at high
frequency levels. By contrast, as prohibitive voice focuses on
pointing out current operational problems and risks, it is likely to
intensify managers’ negative emotions and defensiveness (Liang et
al., 2012; Van Dyne, Cummings, & McLean Parks, 1995; Whiting
et al., 2012), especially at very high frequency levels. Therefore,
we predict that the inverted U-shaped relationship between voice
frequency and managers’ evaluations will be stronger for prohib-
itive voice than for promotive voice.
Hypothesis 1c: The inverted U-shaped relationship between
prohibitive voice and managers’ evaluations is stronger than
that between promotive voice and managers’ evaluations, in
that, at high frequency levels, prohibitive voice is more neg-
atively related to managers’ evaluations than promotive voice
is.
The Moderating Role of LMX
In addition to source characteristics and message content, the
context in which persuasion occurs may also affect its outcomes
(McGuire, 1985). Social persuasion research has suggested that
message receivers’ cognitive evaluations of persuasion endeavors
are influenced largely by the relational context, such as the rela-
tionship quality between the message sender and the message
receiver (e.g., Briñol & Petty, 2009; Menon & Blount, 2003).
People tend to respond more positively to message senders who
belong to their ingroup, because having a shared identity induces
ingroup favoritism and attributions of well-intended motives to
message senders (e.g., Clark & Maass, 1988; Crano & Chen, 1998;
Mason, Conrey, & Smith, 2007), motivates people to process
messages by taking senders’ perspectives (Vaniv & Choshen-
Hillel, 2012), and activates cooperative schemas that facilitate the
messages’ persuasiveness (Tost, Gino, & Larrick, 2012). Such
ingroup favoritism is likely to occur in supervisor–subordinate
dyads with high LMX (e.g., Graen & Uhl-Bien, 1995; Liden,
Sparrowe, & Wayne, 1997). Hence, we cast LMX as an important
indicator of the relational context for voice behavior in the work-
place.
We expect that the curvilinear effect of both promotive voice
and prohibitive voice on managers’ evaluations of voicers is less
salient in supervisor–subordinate dyads with high LMX than those
with low LMX. Because subordinates are treated as ingroup mem-
bers in high LMX contexts (Liden et al., 1997; Uhl-Bien &
Maslyn, 2003), ingroup favoritism tends to cause managers to
make more favorable judgments of their voiced issues, irrespective
of the frequency and content of voice. Prior studies have shown
that when subordinates with high LMX speak up, supervisors are
likely to see their suggestions and even criticisms as assets and as
indicating employees’ initiatives and efforts to contribute to col-
lective interests (e.g., Liden, Wayne, & Stilwell, 1993; Scandura &
Schriesheim, 1994; Uhl-Bien & Maslyn, 2003). In other words, in
high LMX contexts, regardless whether these ingroup subordinates
engage in low or high frequencies of challenging voice and re-
gardless whether the content is promotive or prohibitive, managers
tend to view their voiced issues in a positive light and thus reward
them with more favorable evaluations.
By contrast, in low LMX relationships, formal hierarchical
status is rigidly emphasized, supervisors and subordinates keep a
power distance from each other, and supervisors treat subordinates
as outgroup members (e.g., Liden et al., 1997; Uhl-Bien & Maslyn,
2003). Although managers may generally resist challenging voice
from outgroup members (Menon & Blount, 2003), they may still
appreciate outgroup members who make efforts to initiate changes
to potentially improve organizational practices. Therefore, in low
LMX dyads, promotive/prohibitive voice may be positively asso-
ciated with managers’ evaluations of voicers at low-to-moderate
frequency levels. Yet, managers may not tolerate a high frequency
of promotive/prohibitive voice from these outgroup members.
Very frequent attempts to challenge the status quo from outgroup
members may induce stronger resistance from managers and may
give managers the impression that the voicers put insufficient
effort into considering the voiced issues, causing managers to
dismiss the potentially beneficial outcomes of voice from these
outgroup members. As a result, managers are less likely to reward
outgroup members who voice very frequently by giving them
positive evaluations. In short, we anticipate that in low LMX
dyads, there is a salient inverted U-shape relationship between the
frequency of promotive/prohibitive voice and managers’ evalua-
tions of voicing employees. We thus predict that:
Hypothesis 2a: LMX moderates the curvilinear link between
the frequency of promotive voice and managers’ evaluations
of voicing employees, in that the positive relationship between
 FREQUENCY OF VOICE AND MANAGERS’ RESPONSES
promotive voice and managers’ evaluations is more likely to
become negative at high frequencies of voice in low LMX
dyads than in high LMX dyads.
Hypothesis 2b: LMX moderates the curvilinear link between
the frequency of prohibitive voice and managers’ evaluations
of voicing employees, in that the positive relationship between
prohibitive voice and managers’ evaluations is more likely to
become negative at high frequencies of voice in low LMX
dyads than in high LMX dyads.
So far, we have predicted a stronger inverted U-shaped link
between prohibitive voice and managers’ evaluations than that
between promotive voice and managers’ evaluations (Hypothesis
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
1c), because, at high frequency levels, managers are more likely to
This document is copyrighted by the American Psychological Association or one of its allied publishers.
resist prohibitive voice than promotive voice. We then contended
that because of strong ingroup favoritism, high LMX may have an
overriding positive influence on managers’ evaluations of voicers,
irrespective of frequency levels and voice contents (Hypotheses 2a
and 2b). In high LMX dyads, even for employees who engage in
prohibitive voice behavior frequently, managers may view their
voice in the same positive light as they view promotive voice.
Following this logic, at high frequency levels, high LMX forms a
powerful relational context that buffers the strong negative effects
of prohibitive voice on managers’ evaluations. By contrast, as
managers are more receptive to promotive voice even when it is
voiced very frequently, the buffering effect of LMX for promotive
voice may be relatively weaker. We therefore predict that the
moderating effect of LMX may be stronger for prohibitive voice
than for promotive voice. Taken together, we hypothesize,
Hypothesis 2c: The moderating effect of LMX on the curvi-
linear link between prohibitive voice and managers’ evalua-
tions is stronger than that between promotive voice and man-
agers’ evaluations.
The Mediating Role of Manager-Perceived Voice
Constructiveness
Central to our arguments is the idea that voice frequency, voice
content, and LMX may jointly shape managers’ cognitive assess-
ments of voiced issues and that such cognitive assessments trans-
late into how managers evaluate employees. In this section, we
delve further into this cognitive process. Indeed, the recent devel-
opment of social persuasion theory has highlighted the critical role
of cognitive processes underlying the effects of different types of
persuasion information on message recipients’ attitude formation
and change (Bohner & Dickel, 2011). Albarracín (2002) proposed
the cognition in persuasion model, an extension of social persua-
sion theory, to argue that message characteristics (e.g., voice
content) and peripheral information cues (e.g., voicer characteris-
tics and voicing context) tend to jointly affect message recipients’
assessments of the messages’ beneficial outcomes, which form an
important basis for their favorable or unfavorable reactions to
persuasion. For instance, in their seminal work on cognitive pro-
cesses of social persuasion, Albarracín and Wyer (2001) demon-
strated that although message characteristics, such as the strength
of arguments, tend to induce message receivers’ positive assess-
ments of the messages’ personal and social benefits, persuasion
context (e.g., presence or absence of distraction) can alter these
5
assessments, which in turn influence their subsequent behavioral
reactions (i.e., accepting the messages or not).
Following the same line of reasoning as that of the cognition in
persuasion model, Whiting et al. (2012) contended and demon-
strated that the positive effects of voicer characteristics, voice
content, and voicing context on managers’ evaluations of voicers
are mediated by a key cognitive process, manager-perceived voice
constructiveness—the extent to which managers view the voiced
issues as making positive contributions to the organization. We
thus posit that the frequency of promotive/prohibitive voice and
LMX should interact to generate managers’ judgments either for or
against the constructiveness of voiced issues. When challenging
voice occurs in dyads with high LMX, such a favorable relational
context is likely to cause managers to judge their ingroup employ-
ees’ voice as constructive, irrespective of its frequency level and
content. By contrast, in low LMX dyads, we anticipate that super-
visors’ judgments of voice constructiveness from outgroup mem-
bers is shaped by information cues drawn from subordinates’ voice
frequency and how they frame the content. Hence, following our
earlier arguments, in low LMX contexts, we expect an inverted
U-shaped relationship between frequency of promotive/prohibitive
voice and manager-assessed voice constructiveness. Taken to-
gether, we expect similar quadratic-by-linear interactive effects of
promotive/prohibitive voice frequency and LMX on manager-
perceived voice constructiveness as those on managers’ evalua-
tions, as we predicted in Hypotheses 2a to 2c. We also expect that,
at high frequency levels, LMX may play a more important role in
buffering the negative effect of prohibitive voice than that of
promotive voice on managers’ judgments of the beneficial out-
comes of voiced issues.
Hypothesis 3a: LMX moderates the curvilinear link between
the frequency of promotive voice and manager-perceived
voice constructiveness, in that the positive relationship be-
tween promotive voice and voice constructiveness is more
likely to become negative at high frequencies of voice in low
LMX dyads than in high LMX dyads.
Hypothesis 3b: LMX moderates the curvilinear link between
the frequency of prohibitive voice and manager-perceived
voice constructiveness, in that the positive relationship be-
tween prohibitive voice and voice constructiveness is more
likely to become negative at high frequencies of voice in low
LMX dyads than in high LMX dyads.
Hypothesis 3c: The moderating effect of LMX on the curvi-
linear link between prohibitive voice and manager-perceived
voice constructiveness is stronger than that between promo-
tive voice and manager-perceived voice constructiveness.
According to the cognition in persuasion model, a judgment that
a persuasive message is beneficial is critical in translating persua-
sion effort into attitude change (Albarracín & Wyer, 2001). Indeed,
managers’ favorable cognitive judgment of voiced issues is at the
core of our theoretical reasoning for the joint effect of voicer
characteristic, voice content, and voicing context on managers’
responses. If our key logic is correct, we would expect that
manager-perceived voice constructiveness should mediate the
quadratic-by-linear interactive effects of promotive/prohibitive
voice and LMX managers’ evaluations of voicing employees.
 6 HUANG, XU, HUANG, AND LIU
Hypothesis 4a: LMX moderates the curvilinear indirect effect
of the frequency of promotive voice on managers’ evaluations
of voicing employees, as mediated by manager-perceived
voice constructiveness. This indirect effect is stronger under
conditions of lower than higher LMX.
Hypothesis 4b: LMX moderates the curvilinear indirect effect
of the frequency of prohibitive voice on managers’ evalua-
tions of voicing employees, as mediated by manager-
perceived voice constructiveness. This indirect effect is stron-
ger under conditions of lower than higher LMX.
Hypothesis 4c: LMX has a stronger moderating effect on the
indirect effect of prohibitive voice than on that of promotive
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
voice on managers’ evaluations of voicing employees, as
This document is copyrighted by the American Psychological Association or one of its allied publishers.
mediated by manager-perceived voice constructiveness.
Overview of Studies
We conducted three studies to progressively test our model. In
Study 1, using a Chinese sample, we tested the quadratic-by-linear
interactive effects of promotive/prohibitive voice frequency and
LMX on managers’ evaluations of voicers (Hypotheses 1a–1c and
2a–2c). In Study 2, to address potential for common source bias
(i.e., voice and managers’ evaluations were all rated by managers)
and the cross-cultural generalizability of our model, we replicated
Study 1 by asking employees rather than supervisors to report
promotive/prohibitive voice frequencies and by using a sample
collected from the United States. In Study 3, using another U.S.
sample, we examined the complete model by testing the mediating
role of voice constructiveness. In the voice literature, managerial
evaluations of voicing employees have been widely used to reflect
the extent to which managers reward voice behaviors. This con-
struct has been operationalized in terms of manager-assessed em-
ployee promotability and their evaluations of employees’ overall
performance (Burris, 2012; Burris et al., 2013; Maynes & Podsa-
koff, 2014; Whiting et al., 2008, 2012). We therefore followed this
literature to examine both promotability (Studies 1 and 2) and
manager-rated overall performance evaluations (Study 3).
Study 1
Method
Sample and procedure. We conducted Study 1 in a state-
owned commercial bank in a northern city of China.1 This study
received approval from the ethics committee with the project code
592913 (Back Stabbing and Supervisory Retaliation: Conse-
quences of Voice) at The Hong Kong Polytechnic University.
Respondents were front-line employees and their supervising man-
agers from the retail banking and the counter service in different
branches of the bank. Their main job responsibility was to provide
customers with financial services, such as transactional accounts,
personal loans, and mortgages. HR managers of the bank helped us
invite all employees and immediate supervisors from 31 branches
to participate in our survey. And all these employees and super-
visors were informed that participation was voluntary and confi-
dentiality was ensured. Each supervisor supervised three to seven
employees. The last author and his assistants conducted the pencil
and paper surveys in each branch in two time waves. At Time 1,
supervising managers were asked to report employees’ promotive
voice and prohibitive voice, and employees were asked to rate
LMX with their supervising managers. We also collected data on
the control variables at Time 1. When we executed the surveys
during working hours, subordinates were asked to gather in a
meeting room in groups. They received a questionnaire, a return
envelope, and a letter of introduction. Their immediate supervisors
were asked to answer a separate questionnaire in another meeting
room. To ensure confidentiality, respondents were instructed to
seal the completed questionnaires in the return envelopes and
return them directly to the researchers on site. At Time 2, 3 months
later, managers were asked to evaluate employees’ promotability.
We distributed surveys to 164 employees and 31 supervising
managers. The final sample consisted of 147 employees and 31
managers. The effective response rates were approximately 90%
for employees and 100% for managers. For the employee sample,
39.6% were male, and 95.9% had received a college education or
above. The mean age and organizational tenure were 27.7 years
and 3 years, respectively. For the manager sample, 33.6% were
male, and 93.2% had received a college education or above. The
mean age and organizational tenure were 35.3 years and 6.6 years,
respectively.
Measures. All measures used in this study were developed
originally in English. These measures were translated into Chinese
and back-translated into English by bilingual experts. The back-
translated English version was compared with the original English
version for equivalence and agreement (Brislin, Lonner, &
Thorndike, 1973). Promotive voice and prohibitive voice were
each measured using a five-item, 7-point scale developed by Liang
et al. (2012). A sample item for promotive voice is, “This subor-
dinate raises suggestions to improve the unit’s working proce-
dure.” A sample item for prohibitive voice is, “This subordinate
speaks up honestly with problems that might cause serious loss to
the work unit, even when/though dissenting opinions exist” (1 ⫽
never, 7 ⫽ very often; ␣ ⫽ .97 and ⫽ .92, respectively). Subor-
dinates assessed LMX using the 5-point LMX-7 scale (Graen &
Uhl-Bien, 1995). A sample item is, “How would you characterize
your working relationship with your leader?” (1 ⫽ extremely
ineffective, 5 ⫽ extremely effective; ␣ ⫽ .91). In Study 1, we
captured managers’ evaluations of voicing employees in the form
of manager-rated promotability of the employees. We assessed
promotability using a three-item, seven-point scale adapted from
Burris (2012; 1 ⫽ strongly disagree, 7 ⫽ strongly agree; ␣ ⫽ .81).
We used two items directly from Burris’s measure (“If a position
were available, I would recommend this person for a promotion”
and “If this person was promoted, I would expect him to perform
in his new position”). Because the third item could not fully
capture the construct of promotability (“How would you rate this
person’s performance based on what you know”), we replaced this
third item using a modified one in Chinese. This modified item
now reads, “This employee has greater potential to be promoted in
the future.”
1
The data presented in this study were part of a broader data collection
effort. Although the data of our Study 1 have been used by Wu Liu to
develop a paper on “lateral voice” in another paper (Liu, Tangirala, Lam,
Chen, Jia, & Huang, 2015), the variables used in this lateral voice paper do
not overlap with the variables used in our Study 1 at all.
 FREQUENCY OF VOICE AND MANAGERS’ RESPONSES
Following previous studies (e.g., Burris, 2012), we controlled
for the effects of demographic variables of subordinates (i.e., age,
gender, education, and organizational tenure). We also controlled
for the effect of employees’ past performance, using Motowidlo
and Van Scotter’s (1994) three-item scale, to rule out the halo
effect on supervisors’ evaluations of employees’ promotability
(␣ ⫽ .97). In addition, past studies have suggested that managers
tend to attribute low performers’ proactive behavior as being
driven by impression management motives (De Stobbeleir, Ash-
ford, & de Luque, 2010). Therefore, we controlled for supervisors’
attribution of employees’ voice to impression management mo-
tives by using a two-item scale adapted from Lam, Huang, and
Snape (2007; ␣ ⫽ .73).2
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Results
Before testing our hypotheses, we performed a series of confir-
matory factor analyses (CFAs) to examine the measurement
model, which included promotive voice, prohibitive voice, LMX,
and promotability. Results suggested that the four-factor measure-
ment model yielded a better model fit (comparative fit index
[CFI] ⫽ .98, incremental fit index [IFI] ⫽ .98, Tucker-Lewis
index [TLI] ⫽ .98, root mean square error of approximation
[RMSEA] ⫽ .05) than the three-factor model that combined both
types of voice into one factor (CFI ⫽ .89, IFI ⫽ .89, TLI ⫽ .88,
RMSEA ⫽ .11), with a significant chi-square difference
(⌬␹2(3) ⫽ 273.91); or the two-factor model with two types of
voice as one factor, and LMX and promotability combined as the
other factor (CFI ⫽ .78, IFI ⫽ .78, TLI ⫽ .75, RMSEA ⫽ .16;
⌬␹2(5) ⫽ 518.38); or finally, the single-factor model (CFI ⫽ .53,
IFI ⫽ .54, TLI ⫽ .47, RMSEA ⫽ .24; ⌬␹2(6) ⫽ 1142.25).
Descriptive statistics and correlations for the study variables are
presented in Table 1. The sample of Study 1 had a nested data
structure: Employees were nested within supervisors of their work
groups. We calculated ICC1 from a one-way random-effects
ANOVA model with Bartko’s (1976) formula. The ICC1 for
promotability was .27 (p ⬍ .001), suggesting substantial variance
of promotability at the group level. Therefore, we used multilevel
analysis to test the model to control for between-groups variances
(Raudenbush & Bryk, 2002). We centered the independent vari-
able at its grand means before evaluating the regression equations
(cf. Aiken & West, 1991; Cohen, Cohen, West, & Aiken, 2003).
And we used this centering procedure in all three studies.
Hypotheses 1a and 1b predicted that the frequencies of promo-
tive and prohibitive voice, respectively, have inverted U-shaped
relationships with managers’ evaluations of voicing employees. As
shown in Model 1 of Table 2, we entered the control variables in
the first step. Employee gender (␤ ⫽ ⫺.30, p ⬍ .05), employee
past performance (␤ ⫽ .76, p ⬍ .001), and manager’s attribution
(␤ ⫽ ⫺.15, p ⬍ .05) were significantly related to manager-rated
promotability. In Model 2, we entered promotive voice and pro-
hibitive voice, which were not significantly related to promotabil-
ity. In Model 3, we entered the quadratic terms of promotive voice
and prohibitive voice and found that neither promotive voice nor
prohibitive voice had a nonlinear relationship with promotability.
Thus, Hypotheses 1a and 1b were not supported. Hypothesis 1c
predicted that, at high frequency levels, the negative effect of
prohibitive voice is stronger than that of promotive voice. As we
7
did not find curvilinear main effects of either promotive voice or
prohibitive voice, Hypothesis 1c was not supported.
Next, we tested the moderating effect of LMX on the curvilinear
relationship between the frequency of promotive/prohibitive voice
and managers’ evaluations of voicing employees (Hypotheses 2a
and 2b). As presented in Table 2, we entered the moderator LMX
in Model 4 and then the linear interaction of voice frequency
(promotive and prohibitive voice) and LMX in Model 5. None of
these terms was significant for promotability. In Model 6, the
interaction term of quadratic promotive voice and LMX was not
significant (␤ ⫽ ⫺.01, ns), and thus Hypothesis 2a was not
supported. Yet, the interaction term of quadratic prohibitive voice
and LMX was significant (␤ ⫽ .32, p ⬍ .01, ⌬R2 ⫽ .15). We thus
performed additional analyses to test Hypothesis 2b by examining
the simple slopes of the regression lines corresponding to all
possible combinations of different frequency levels of prohibitive
voice with high and low levels of LMX (cf. Aiken & West, 1991).
Results of simple slope tests (see Table 3) showed that, in the case
of high LMX, the simple slope of the regression lines was signif-
icant for promotability only at a very high frequency level (␤ ⫽
.97, p ⬍ .05, 2 SD above the mean). By contrast, in the case of low
LMX, the very low (␤ ⫽ 1.79, p ⬍ .01, 2 SD below the mean) and
low frequencies of prohibitive voice (␤ ⫽ .96, p ⬍ .01, 1 SD below
the mean) were positively related to promotability; the medium
(␤ ⫽ ⫺.14, ns) and high (␤ ⫽ ⫺.69, ns, 1 SD above the mean)
voice frequencies were not significantly related to promotability;
and the very high frequency of prohibitive voice (␤ ⫽ ⫺.85, p ⬍
.05, 2 SD above the mean) was negatively related to promotability.
To facilitate the interpretation of this quadratic-by-linear interac-
tion effect, we visualize the interaction in Figure 2A. As shown in
the figure, when LMX was high, the effect of prohibitive voice
became positive rather than negative at the very high frequency
level. When LMX was low, there was an inverted U-shaped
relationship between the frequency of prohibitive voice and pro-
motability. Thus, Hypothesis 2b was supported.
Hypothesis 2c predicts that the moderating effect of LMX on the
curvilinear linkage between prohibitive voice and managers’ eval-
uations is stronger than that between promotive voice and manag-
ers’ evaluations. To compare the moderating effects of LMX on
these two linkages, we performed a relative weight analysis (John-
son, 2000) by using RWA-Web (Tonidandel & LeBreton, 2015).
Confidence intervals (95% CIs) for the relative weight of each
interactive effect and all corresponding significance tests were
based on bootstrapping with 10,000 replications. The results indi-
cated that the quadratic-by-linear effect of prohibitive voice and
LMX was the most salient predictor, explaining roughly 35% of
the variance of promotability (relative weight [RW] ⫽ .35 [.02,
.15]). Moreover, a relative comparison analysis showed that the
relative weight for the quadratic-by-linear interactive effect of
2
To examine the robustness of our findings, we controlled for different
sets of variables across the three studies. We performed additional tests to
see whether we would obtain different results by removing the control
variables in the three studies. If we removed all the control variables, we
obtained essentially the same results in all three studies. Also, as prior
employee performance is strongly related to managers’ responses to em-
ployee voice across all three studies, we did another set of analyses by
controlling for prior performance only. Again, we found essentially the
same results in all three studies.
 8 HUANG, XU, HUANG, AND LIU

Table 1
Means, Standard Deviations, and Correlations Among the Variables (Study 1)

Variables M SD 1 2 3 4 5 6 7 8 9 10
a
1. Employee age 27.68 3.91
2. Employee genderb .60 .49 .09
3. Employee educationc .96 .20 ⫺.02 ⫺.03
4. Employee tenurea 3.02 2.32 .41ⴱⴱⴱ .17ⴱ ⫺.03
5. Past performance 6.07 .87 ⫺.14 .08 ⫺.07 ⫺.10
6. Manager’s attribution 3.16 1.41 ⫺.03 ⫺.17ⴱ .01 ⫺.06 ⫺.29ⴱⴱⴱ
7. Promotive voice 3.96 1.52 .17ⴱ .15 .07 .19ⴱ .24ⴱⴱ .10
8. Prohibitive voice 3.64 1.38 .12 .08 .01 .20ⴱ .23ⴱ .06 .72ⴱⴱⴱ
9. LMX 3.67 .69 ⫺.05 ⫺.05 .07 ⫺.11 .17 ⫺.02 .07 ⫺.11
10. Promotability 5.51 1.08 ⫺.02 .05 ⫺.07 ⫺.11 .66ⴱⴱⴱ ⫺.36ⴱⴱⴱ .19ⴱ .21ⴱ .17ⴱ
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Note. N ⫽ 147. LMX ⫽ leader-member exchange quality.

Age and tenure were measured in years. b 0 ⫽ male; 1 ⫽ female. 0 ⫽ high school or below; 1 ⫽ college or above.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

a c
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.

prohibitive voice and LMX (RW ⫽ .35) was significantly different
from the relative weight for the quadratic-by-linear interactive
effect of promotive voice and LMX (RW ⫽ .05 [⫺.13, ⫺.02]).
Therefore, Hypothesis 2c was supported.
Discussion
In summary, in Study 1 we did not find any significant main
curvilinear effects of the frequency of promotive and prohibitive
voice on manager-rated promotability (Hypotheses 1a, 1b, and 1c),
but we did find support for Hypotheses 2b and 2c. Specifically, we
found robust results for the inverted U-shaped link between voice
frequency and promotability under the condition of low rather than
high LMX, and for prohibitive voice rather than promotive voice.
Study 1 had two weaknesses, however. First, the independent
variable and the outcome variable were both rated by supervisors,
leading to a concern about common source bias, even though we
used a time-lagged design. Second, we collected data from China,
which has a large power distance culture (Hofstede, 2001) that
may predispose employees to remain silent in organizations and
managers to be less tolerant of employees’ challenging voice
(Huang, Van de Vliert, & Van der Vegt, 2005). We therefore
conducted Study 2 to address these two concerns.
Study 2
Method
Sample and procedure. We followed the same procedure as in
Study 1 to conduct a two-wave survey with a 4-week lag from a large
information technology company in the United States. But unlike in
Study 1, in which we asked supervisors to report employee voice, in
Study 2 we asked employees to report promotive voice and prohibi-
tive voice to address potential common method bias. This study was
conducted under Institutional Review Board Protocol #14286 (Inves-

Table 2
Hierarchical Multilevel Analyses for Promotability (Study 1)

Promotability
Variables M1 M2 M3 M4 M5 M6

Employee age .00 (.02) .00 (.02) .00 (.01) .00 (.02) .00 (.02) ⫺.00 (.02)
Employee gender ⫺.30ⴱ (.15) ⫺.30ⴱ (.15) ⫺.29 (.15) ⫺.28 (.15) ⫺.26 (.15) ⫺.28 (.15)
Employee education .19 (.36) .10 (.36) .10 (.36) .10 (.36) .19 (.36) .14 (.35)
Employee tenure .00 (.00) .00 (.00) .00 (.00) .00 (.00) .00 (.00) .00 (.00)
Past performance .76ⴱⴱⴱ (.11) .71ⴱⴱⴱ (.11) .71ⴱⴱⴱ (.11) .71ⴱⴱⴱ (.11) .71ⴱⴱⴱ (.11) .68ⴱⴱⴱ (.10)
Manager’s attribution ⫺.15ⴱ (.07) ⫺.17 (.13) ⫺.17ⴱ (.07) ⫺.17ⴱ (.07) ⫺.17ⴱ (.07) ⫺.18ⴱⴱ (.06)
Promotive voice (PMV) .00 (.13) ⫺.00 (.13) ⫺.01 (.13) .01 (.14) .05 (.14)
Prohibitive voice (PHV) .18 (.13) .19 (.13) .19 (.13) .21 (.14) .16 (.13)
PMV2 .04 (.10) .03 (.11) .05 (.11) .11 (.11)
PHV2 .03 (.09) .03 (.09) .03 (.09) ⫺.08 (.09)
LMX .01 (.08) ⫺.02 (.08) ⫺.23ⴱ (.11)
PMV ⫻ LMX ⫺.02 (.14) ⫺.10 (.14)
PHV ⫻ LMX ⫺.13 (.12) .05 (.13)
PMV2 ⫻ LMX ⫺.01 (.12)
PHV2 ⫻ LMX .32ⴱⴱ (.10)
⌬␹2(df) 52.34 (6)ⴱⴱⴱ 3.88 (2) .58 (2) .03 (1) 2.54 (2) 9.62 (2)ⴱⴱ
Pseudo ⌬R2 .49 .05 .01 .00 .01 .15
Note. M ⫽ Model; LMX ⫽ leader-member exchange quality.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
 FREQUENCY OF VOICE AND MANAGERS’ RESPONSES
Table 3
Tests of Simple Slopes for Quadratic-By-Linear Interactions of Prohibitive Voice and LMX for
Promotability (Studies 1 and 2)
Moderator X (2 SD low) X (1 SD low)
Study 1
High LMX ⫺.67 ⫺.24
Low LMX 1.79ⴱⴱ .96ⴱⴱ
Study 2
High LMX .41 .42ⴱ
Low LMX .85 .48
Note. X ⫽ independent variable; LMX ⫽ leader-member exchange quality.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
tigating Manager-Subordinate Relationships from the Manager Lens)
at University of Nebraska-Lincoln. We first obtained a letter of
support from the Company’s Chief Operating Officer and then invited
participants to complete paper-and-pencil surveys during working
hours. All invited participants were briefed about the purpose and
procedures of this survey study. And all these employees and super-
visors were informed that participation was voluntary and confiden-
tiality was ensured. On average, the managers’ span of control was
about eight people (ranging from four to14). Individuals were nested
in teams which typically specialized in the same functional area, or
similar/relevant areas. We invited 779 employees and 97 supervising
managers from the same location to participate in our survey. Re-
spondents were from different functional teams (e.g., business oper-
ations, data service consulting, cloud operations, small business so-
lutions, mobile solutions, global sales, enterprise application services,
and brand communication). The final sample consisted of 289 em-
ployees and 48 managers. The effective response rates were approx-
imately 37.1% for employees and 49.5% for managers. For the
employee sample, 40.8% were male, and 25.6% held a master’s
degree or above. The mean age and organizational tenure of the
employees were 29.8 and 4.6 years, respectively. For the manager
sample, 70.9% were male, and all manager respondents held a bach-
elor’s degree or above. The mean age and organizational tenure of the
manager sample were 38.2 and 11.7 years, respectively.
Measures. We used the same measures as those in Study 1,
except that we used a different measure for promotability. We used
only two items from Burris’ (2012) measure of promotability and
removed the third item, which focused on overall performance. The
Cronbach’s alpha coefficients was .81 for promotive voice, .87 for
prohibitive voice, .88 for LMX, and .92 for promotability. Like in
Study 1, we controlled for employees’ demographic variables and
past performance (␣ ⫽ .91). We did not control for impression
management because we were not allowed to add additional variables
into our questionnaires and also impression management did not
substantially affect our results.
Results
We performed a series of CFAs to examine the measurement
model. Similar to Study 1, the results suggested that the four-factor
measurement model yielded a better model fit (CFI ⫽ .99, IFI ⫽
.99, TLI ⫽ .99, RMSEA ⫽ .03) than the three-factor model
(CFI ⫽ .83, IFI ⫽ .83, TLI ⫽ .81, RMSEA ⫽ .10), with a
9
␤
X (medium) X (1 SD high) X (2 SD high)
.18 .53 .97ⴱ
.14 ⫺.69 ⫺.85ⴱ
.43 .44ⴱ .44
.11 ⫺.27ⴱ ⫺.69ⴱⴱ
significant chi-square difference (⌬␹2(3) ⫽ 602.12); or the two-
factor model (CFI ⫽ .54, IFI ⫽ .54, TLI ⫽ .48, RMSEA ⫽ .17;
⌬␹2(5) ⫽ 1335.00); or finally, the single-factor model (CFI ⫽ .37,
IFI ⫽ .37, TLI ⫽ .29, RMSEA ⫽ .20; ⌬␹2(6) ⫽ 1847.69).
Descriptive statistics and correlations for the variables are pre-
sented in Table 4. This sample also had a nonindependent data
structure. The ICC1 for promotability was .30 (p ⬍ .001). There-
fore, we used multilevel analysis to test the model.
We followed the same procedure as that in Study 1 to test the
hypotheses. As shown in Table 5 (Model 3), promotive voice did
not have a significant nonlinear effect on promotability. Therefore,
Hypothesis 1a was not supported. In contrast, the quadratic term of
prohibitive voice was negatively related to promotability
(␤ ⫽ ⫺.17, p ⬍ .01). The simple slopes of the regression line for
promotability at the very low, low, medium, high, and very high
frequency levels of prohibitive voice (␤ ⫽ 1.29, p ⬍ .001; ␤ ⫽ .82,
p ⬍ .001; ␤ ⫽ .34, p ⬍ .001; ␤ ⫽ ⫺.14, ns; ␤ ⫽ ⫺.61, p ⬍ .001)
indicated an inverted U-shaped relationship. These findings lent
support for Hypothesis 1b. The fact that we found a significant
inverted U-shaped relationship for prohibitive voice but not for
promotive voice also lent initial support for Hypothesis 1c.
As shown in Model 6 of Table 5, we did not find a significant
quadratic-by-linear interactive effect of promotive voice and LMX
on promotability. Thus, Hypothesis 2a was not supported. We did
find, however, that the quadratic-by-linear interaction of prohibi-
tive voice and LMX was significant for promotability (␤ ⫽ .09,
p ⬍ .05, ⌬R2 ⫽ .04). As shown in Table 3, simple slope tests found
a significant inverted U-shaped link between prohibitive voice and
promotability in low rather than high LMX dyads (also see Figure
2B). These findings again supported Hypothesis 2b. Also, LMX
was found to moderate the curvilinear effect of prohibitive voice
rather than that of promotive voice on managers’ evaluations,
which lent initial support to Hypothesis 2c. We then performed a
relative importance test by using RWA-Web (Tonidandel &
LeBreton, 2015), as we did in Study 1, to further confirm Hypoth-
eses 1c and 2c. Results indicated that the quadratic effect of
prohibitive voice explained more variance of promotability
(RW ⫽ .15) than the quadratic effect of promotive voice did
(RW ⫽ .03 [⫺.08, ⫺.01]), thereby supporting Hypothesis 1c.
Moreover, the quadratic-by-linear interactive effect of prohibitive
voice and LMX (RW ⫽ .22) explained significantly more variance
of promotability than the quadratic-by-linear interactive effect of
 10 HUANG, XU, HUANG, AND LIU

A: Study 1
7

Promotability
5
High LMX

Low LMX
4

2
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

1
Prohibitive voice
Low High

B: Study 2

7
6.5
6
Promotability

5.5
5 High LMX

4.5 Low LMX

4
3.5
3
2.5
2
Prohibitive voice
Low High

Figure 2. Results of the quadratic-by-linear interactive effect of prohibitive voice and leader-member exchange
quality (LMX) on promotability (Studies 1 and 2).

promotive voice and LMX did (RW ⫽ .06 [⫺.09, ⫺.02]), lending
support to Hypothesis 2c.
Discussion
Using employee-rated voice behavior in Study 2, we not only
replicated the significant results reported in Study 1 that supported
Hypotheses 2b and 2c, but also found new evidence to support
Hypotheses 1b and 1c. The key limitation of Studies 1 and 2,
however, is that we did not examine the mediating effect of
manager-perceived voice constructiveness. We therefore con-
ducted Study 3 to test this mechanism using another sample.
Also, in Study 3, we attempted to address three additional
methodological limitations of Studies 1 and 2. First, the differ-
ences in the results of Studies 1 and 2 could have been caused by
either cultural differences (China vs. the United States) or differ-
ences in how voice was measured (manager-rated vs. employee-
rated voice). Hence, in Study 3, using another U.S. sample (hold-
ing national culture constant), we measured challenging voice
using manager-rated voice, which is more in line with our theory:
Managers should be aware of the frequency of employees’ chal-
lenging voice. Second, in Studies 1 and 2, we examined manager-
rated promotability as the indicator of managers’ evaluations.
Although promotability may reflect managers’ evaluations of em-
ployees, it is important to examine whether our model can be
extended to predict managers’ evaluations of employees’ overall
performance, because past studies have used overall performance
as another major indicator of managers’ evaluations of employees
(Burris, 2012; Burris et al., 2013; Maynes & Podsakoff, 2014;
Whiting et al., 2008, 2012). Hence, in Study 3, we used manager-
rated overall performance of voicing employees as the dependent
variable. Third, one possible criticism of Studies 1 and 2 is that the
results may have been confounded with some individual difference
variables of both subordinates and supervisors, such as subordi-
nates’ proactivity and managers’ negative affectivity and openness
 FREQUENCY OF VOICE AND MANAGERS’ RESPONSES 11

Table 4
Means, Standard Deviations, and Correlations Among the Variables (Study 2)

Variables M SD 1 2 3 4 5 6 7 8 9
a
1. Employee age 29.80 3.50
2. Employee genderb .60 .53 .00
3. Employee educationc .26 .44 ⫺.02 ⫺.02
4. Employee tenurea 4.56 2.95 .69ⴱⴱⴱ .03 ⫺.10
5. Past performance 4.92 .84 .02 ⫺.10 .01 ⫺.02
6. Promotive voice 3.68 .69 ⫺.06 .02 ⫺.01 ⫺.10 .35ⴱⴱⴱ
7. Prohibitive voice 3.65 .83 .02 .04 ⫺.08 ⫺.04 .37ⴱⴱⴱ .13ⴱ
8. LMX 3.68 .71 .04 ⫺.10 .07 .10 .34ⴱⴱⴱ .15ⴱ .04
9. Promotability 4.93 1.31 .06 ⫺.11 .01 .03 .51ⴱⴱⴱ .24ⴱⴱⴱ .32ⴱⴱ .31ⴱⴱⴱ
Note. N ⫽ 289. LMX ⫽ leader-member exchange quality.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

a
Age and tenure were measured in years. b 0 ⫽ male; 1 ⫽ female. c
0 ⫽ bachelor’s degree or below; 1 ⫽ master’s degree or above.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

to voice. In Study 3, we tested our model by controlling additional
individual difference variables.
Study 3
Method
Sample and procedure. We invited employees and their su-
pervising managers working at a large financial services company
in the United States to participate in this study. This study was
conducted under Institutional Review Board Protocol #16 – 438
EP1703 (The Impact of Prosocial Behaviors on Workplace Out-
comes) at Auburn University. We first obtained a letter of support
from the company’s Human Resources Department and then in-
vited participants to complete online surveys. All invited partici-
pants were briefed about the purpose and procedures of this survey
study. In addition, all these employees and supervisors were in-
formed that participation was voluntary and confidentiality was
ensured. The managers’ average span of control was 8.8 people,
ranging from five to 14. Individuals were nested in teams special-
izing in the same functional area, or similar/relevant areas. The
invited participants worked in a number of areas, such as home and
auto insurance, corporate finance, asset management, financial
planning, investment accounting, operations, and communications.
Survey data were collected in one location at two time points that
were separated by approximately four weeks. At Time 1, we
invited 565 employees to answer questions about their proactive
personality and LMX. We received 392 completed questionnaires
(69.4%). We then asked the 63 supervising managers of these 392
employee respondents to answer questions about their subordi-
nates’ past performance, promotive voice, prohibitive voice, and
voice constructiveness, as well as their own negative affect and
openness. We collected 206 matched questionnaires from 46 su-
pervising managers (73.0%). At Time 2, we asked those 46 man-
ager respondents to rate their subordinates’ overall performance.
All 46 managers returned their questionnaires (100%). Therefore,
our final sample consisted of 206 employees nested under 46
supervising managers. Of these 206 employees, 47.6% were fe-
male and 28.2% held a master’s degree or above. The average age

Table 5
Hierarchical Multilevel Analyses for Promotability (Study 2)

Promotability
Variables M1 M2 M3 M4 M5 M6

Employee age .01 (.02) .01 (.02) .01 (.02) .02 (.02) .01 (.02) .02 (.02)
Employee gender ⫺.13 (.11) ⫺.16 (.11) ⫺.14 (.11) ⫺.11 (.10) ⫺.12 (.10) ⫺.13 (.10)
Employee education .04 (.09) .08 (.08) .10 (.08) .08 (.08) .09 (.08) .10 (.08)
Employee tenure ⫺.02 (.03) ⫺.01 (.03) ⫺.01 (.03) ⫺.02 (.03) ⫺.02 (.03) ⫺.01 (.03)
Past performance .93ⴱⴱⴱ (.08) .85ⴱⴱⴱ (.08) .82ⴱⴱⴱ (.08) .74ⴱⴱⴱ (.08) .75ⴱⴱⴱ (.08) .72ⴱⴱⴱ (.08)
Promotive voice (PMV) .06 (.06) .04 (.06) .04 (.06) .03 (.06) .07 (.06)
Prohibitive voice (PHV) .25ⴱⴱⴱ (.07) .19ⴱⴱ (.07) .23ⴱⴱⴱ (.07) .27ⴱⴱⴱ (.07) .27ⴱⴱⴱ (.07)
PMV2 ⫺.04 (.04) ⫺.04 (.04) ⫺.05(.04) ⫺.05 (.04)
PHV2 ⫺.17ⴱⴱ (.05) ⫺.11ⴱ (.05) ⫺.15ⴱⴱ (.06) ⫺.09ⴱ (.06)
LMX .28ⴱⴱⴱ (.07) .23ⴱⴱⴱ (.07) .07 (.10)
PMV ⫻ LMX ⫺.01 (.06) ⫺.01 (.06)
PHV ⫻ LMX .12ⴱ (.05) .16ⴱⴱ (.05)
PMV2 ⫻ LMX ⫺.00 (.04)
PHV2 ⫻ LMX .09ⴱ (.04)
⌬␹2 (df) 112.9 (5)ⴱⴱⴱ 14.23 (2)ⴱⴱⴱ 10.84 (2)ⴱⴱ 16.42 (1)ⴱⴱⴱ 5.65 (2) 5.08 (2)
Pseudo ⌬R2 .26 .02 .02 .01 .03 .04
Note. M ⫽ Model. LMX ⫽ leader-member exchange quality.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
 12 HUANG, XU, HUANG, AND LIU

and organizational tenure were 34.3 and 5.6 years, respectively.
The average dyadic tenure was 3.6 years. For the manager sample,
30.4% were female, and all of them held a master’s degree or
above. The average age and organizational tenure were 41.6 and
12.9 years, respectively.
Measures. Promotive voice (␣ ⫽ .91), prohibitive voice (␣ ⫽
.92), and LMX (␣ ⫽ .96) were all measured using the same scales
as those used in Studies 1 and 2. Voice constructiveness was
adopted from Whiting et al.’s (2012) measure, which originated
from Gorden (1988). The items are, “This employee’s suggestions/
comments are likely to enhance the performance of his/her work
group” and “This employee’s suggestions/comments are construc-
tive” (␣ ⫽ .84; 1 ⫽ strongly disagree, 7 ⫽ strongly agree). Overall
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
TLI ⫽ .95, RMSEA ⫽ .06) than the four-factor model that
combined both types of voice into one factor (CFI ⫽ .80, IFI ⫽
.80, TLI ⫽ .78, RMSEA ⫽ .12), with a significant chi-square
difference (⌬␹2(4) ⫽ 480.23); or the four-factor model that com-
bined voice constructiveness and overall performance into one
factor (CFI ⫽ .90, IFI ⫽ .90, TLI ⫽ .87, RMSEA ⫽ .09; ⌬␹2(4) ⫽
166.81); or the three-factor model with the two types of challeng-
ing voice combined into one factor, voice constructiveness and
overall performance combined into one factor, and LMX as one
factor (CFI ⫽ .75, IFI ⫽ .75, TLI ⫽ .72, RMSEA ⫽ .14; ⌬␹2(7) ⫽
647.03); or the two-factor model with the two types of voice as one
factor and all the others as the other factor (CFI ⫽ .81, IFI ⫽ .81,
TLI ⫽ .79, RMSEA ⫽ .12; ⌬␹2(9) ⫽ 453.73); or finally, the

performance evaluation was measured using a three-item scale single-factor model (CFI ⫽ .48, IFI ⫽ .48, TLI ⫽ .42, RMSEA ⫽
This document is copyrighted by the American Psychological Association or one of its allied publishers.

developed by MacKenzie, Podsakoff, and Fetter (1991). A sample
item is, “All things considered, this employee performs his/her job
the way I like to see it performed” (␣ ⫽ .88; 1 ⫽ strongly disagree,
7 ⫽ strongly agree).
We also controlled for a number of variables in addition to the
demographic variables. First, we controlled for employees’ proac-
tive personality (six-item scale, ␣ ⫽ .89; Parker, 1998), which has
been found to significantly predict employee voice (Detert &
Burris, 2007; Seibert et al., 2001). Second, we controlled for
managers’ ratings of employees’ past performance (three-item
scale, ␣ ⫽ .89; MacKenzie et al., 1991), managers’ negative
affectivity (five-item scale, ␣ ⫽ .82; Thompson, 2007), and man-
agers’ openness (two-item scale, ␣ ⫽ .88; Gosling, Rentfrow, &
Swann, 2003), which were all found to be significantly related to
managers’ evaluations of employees (Detert & Burris, 2007; Tan-
girala & Ramanujam, 2012).
Results
We performed a series of CFAs to examine the measurement
model, which included promotive voice, prohibitive voice, LMX,
manager-perceived voice constructiveness, and manager-rated
overall performance. Results suggested that the five-factor mea-
surement model yielded a better model fit (CFI ⫽ .96, IFI ⫽ .96,
.19; ⌬␹2(10) ⫽ 647.03). Descriptive statistics and correlations for
the study variables are presented in Table 6. Again, the sample had
a nonindependent data structure. The ICC1 was .18 for manager-
rated overall performance (p ⬍ .01). Therefore, we employed
multilevel analysis to test the model.
We first tested Hypotheses 3a and 3b, which predicted
quadratic-by-linear interactive effects of the two types of challeng-
ing voice and LMX on the mediator: manager-perceived voice
constructiveness. As shown in Table 7 (Model 6), we found a
nonsignificant quadratic-by-linear interactive effect of promotive
voice and LMX, but a significant quadratic-by-linear interactive
effect of prohibitive voice and LMX on voice constructiveness
(␤ ⫽ .34, p ⬍ .001, ⌬R2 ⫽ .09). Simple slope tests presented in
Table 8 and the plot of this interaction in Figure 3A showed a
similar pattern of interactions as those found in Studies 1 and 2.
Hence, although Hypothesis 3a was not supported, Hypothesis 3b
was supported. We then performed a relative importance test by
using RWA-Web (Tonidandel & LeBreton, 2015) to test Hypoth-
esis 3c. The results revealed that the quadratic-by-linear interactive
effect of prohibitive voice and LMX explained significantly more
variance of voice constructiveness (RW ⫽ .70) than the quadratic-
by-linear interactive effect of promotive voice and LMX did
(RW ⫽ .23 [⫺.19, ⫺.01]), thus supporting Hypothesis 3c.

Table 6
Means, Standard Deviations, and Correlations Among the Variables (Study 3)

Variables M SD 1 2 3 4 5 6 7 8 9 10 11 12 13 14
a
1. Employee age 34.33 4.79
2. Employee genderb .48 .50 ⫺.04
3. Employee educationc .28 .45 .01 .00
4. Employee tenurea 5.57 4.17 .49ⴱⴱⴱ ⫺.08 ⫺.00
5. Dyadic tenure a
3.56 2.12 .34ⴱⴱⴱ ⫺.10 ⫺.06 .80ⴱⴱⴱ
6. Proactivity 4.70 1.09 .08 ⫺.04 .00 ⫺.04 ⫺.03
7. Past performance 4.23 1.38 ⫺.03 .01 .06 ⫺.01 ⫺.01 .05
8. Negative affectivity 3.29 .98 ⫺.03 ⫺.13 ⫺.01 .13 .04 ⫺.04 ⫺.12
9. Managerial openness 4.02 .50 ⫺.01 .04 ⫺.03 ⫺.14 ⫺.08 ⫺.03 .13 ⫺.08
10. Promotive voice 4.39 1.14 .06 .04 ⫺.08 .09 .08 .24ⴱⴱⴱ .09 .02 .06
11. Prohibitive voice 4.24 1.27 .06 ⫺.04 .04 .07 .03 .13 .19ⴱⴱ ⫺.03 ⫺.08 .20ⴱⴱ
12. LMX 4.53 1.36 .08 .12 ⫺.06 ⫺.02 ⫺.08 ⫺.10 .07 ⫺.08 ⫺.00 .01 ⫺.00
13. Voice constructiveness 4.41 1.41 ⫺.01 .09 ⫺.02 .00 .07 ⫺.04 .22ⴱⴱ ⫺.06 ⫺.05 .05 .01 .33ⴱⴱⴱ
14. Overall performance 4.50 1.43 ⫺.04 .03 .04 ⫺.05 ⫺.04 ⫺.13 .04 ⫺.11 ⫺.02 .06 .06 .16ⴱ .21ⴱⴱ
Note. N ⫽ 206. LMX ⫽ leader-member exchange quality.
a
Age, tenure, and dyadic tenure were measured in years. b 0 ⫽ male; 1 ⫽ female. c
0 ⫽ bachelor’s degree or below; 1 ⫽ master’s degree or above.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
 FREQUENCY OF VOICE AND MANAGERS’ RESPONSES 13

Table 7
Hierarchical Multilevel Analyses for Manager-Perceived Voice Constructiveness (Study 3)

Voice constructiveness
Variables M1 M2 M3 M4 M5 M6

Employee age .01 (.02) .01 (.02) .00 (.02) ⫺.01 (.02) ⫺.01 (.02) ⫺.01 (.02)
Employee gender .24 (.20) .22 (.19) .223(.19) .16 (.18) .14 (.18) .08 (.17)
Employee education ⫺.07 (.17) ⫺.06 (.17) ⫺.07 (.17) ⫺.02 (.16) ⫺.04 (.16) ⫺.04 (.15)
Employee tenure ⫺.01 (.04) ⫺.01 (.04) .01 (.04) .00 (.04) .01 (.04) .00 (.04)
Dyadic tenure .06 (.08) .05 (.08) .03 (.08) .06 (.07) .06 (.07) .07 (.07)
Past performance .21ⴱⴱ (.07) .21ⴱⴱ (.07) .22ⴱⴱ (.07) .20ⴱⴱ (.07) .16ⴱ (.07) .16ⴱ (.07)
Proactivity ⫺.07 (.10) ⫺.07 (.10) ⫺.08 (.10) ⫺.02 (.09) ⫺.02 (.09) ⫺.00 (.09)
Negative affectivity ⫺.02 (.10) ⫺.03 (.10) ⫺.05 (.10) ⫺.02 (.10) ⫺.04 (.10) ⫺.04 (.09)
Managerial openness ⫺.25 (.20) ⫺.28 (.20) ⫺.30 (.20) ⫺.29 (.19) ⫺.28 (.19) ⫺.31 (.18)
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Promotive voice (PMV) .08 (.11) .07 (.11) .06 (.10) .03 (.10) .07 (.10)
⫺.09 (.10) ⫺.06 (.10) ⫺.04 (.10) ⫺.03 (.10) ⫺.05 (.09)
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Prohibitive voice (PHV)

PMV2 ⫺.05 (.08) ⫺.09 (.08) ⫺.09 (.08) ⫺.13 (.07)
PHV2 ⫺.14 (.09) ⫺.10 (.09) ⫺.09 (.09) ⫺.04 (.08)
LMX .45ⴱⴱⴱ (.09) .45ⴱⴱⴱ (.09) ⫺.01 (.14)
PMV ⫻ LMX .08 (.10) .17 (.10)
PHV ⫻ LMX .17ⴱ (.08) .11(.08)
PMV2 ⫻ LMX .07 (.08)
PHV2 ⫻ LMX .34ⴱⴱⴱ (.08)
⌬␹2(df) 12.10 (9) 1.20 (2) 2.88 (2) 22.37ⴱⴱⴱ (1) 4.91 (2) 17.97ⴱⴱⴱ (2)
Pseudo ⌬R2 .06 .01 .02 .11 .03 .09
Note. M ⫽ Model; LMX ⫽ leader-member exchange quality.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.

We followed the same procedures as those in Studies 1 and 2 to
test the effect of voice frequency on the dependent variable:
manager-rated overall performance. As presented in Model 3 of
Table 9 neither promotive voice nor prohibitive voice had a
nonlinear relationship with overall performance. Thus, Hypotheses
1a, 1b, and 1c were not supported. Also, we did not find a
significant quadratic-by-linear interactive effect of promotive
voice and LMX on overall performance. Thus, Hypothesis 2a was
not supported. This finding also led to the rejection of Hypothesis
4a that predicts a mediating role of voice constructiveness for the
effect of promotive voice. Notably, however, we found a signifi-
cant linear interactive effect of promotive voice and LMX on
overall performance (␤ ⫽ .21, p ⬍ .05; Model 5 of Table 9). This
finding suggests that the positive relationship between promotive
voice frequency and manager-rated overall performance is stron-
ger in high than in low LMX dyads.
As shown in Model 6 of Table 9 we found that the quadratic-
by-linear interaction of prohibitive voice and LMX was significant
(␤ ⫽ .20, p ⬍ .05, ⌬R2 ⫽ .04). Simple slope tests presented in
Table 8 and the plot of this interaction in Figure 3B show an
inverted U-shaped link between frequency of prohibitive voice and
overall performance when LMX is low rather than high, lending
support to Hypothesis 2b. Also, in support of Hypothesis 2c, the
relative importance test by using RWA-Web (Tonidandel & LeB-
reton, 2015) revealed that the quadratic-by-linear interactive effect
of prohibitive voice and LMX explained significantly more vari-
ance of overall performance (RW ⫽ .63) than the quadratic-by-
linear interactive effect of promotive voice and LMX did (RW ⫽
.07 [⫺.18, ⫺.01]).
Next, we tested the effect of voice constructiveness on overall
performance after including the linear and squared terms of voice,
LMX, and the respective interaction terms. As shown in Model 7

Table 8
Tests of Simple Slopes for Quadratic-By-Linear Interactions of Prohibitive Voice and LMX for
Manager-Perceived Voice Constructiveness and Manager-Rated Overall Performance (Study 3)

␤
Moderator X (2 SD low) X (1 SD low) X (medium) X (1 SD high) X (2 SD high)

Voice constructiveness
High LMX ⫺.85 ⫺.38 .09 .56ⴱ 1.02ⴱ
Low LMX 1.36ⴱⴱ .61ⴱ ⫺.14 ⫺.89ⴱⴱⴱ ⫺1.63ⴱⴱ
Overall performance
High LMX ⫺.13 .06 .25 .44 .63
Low LMX 1.12ⴱ .57ⴱ .02 ⫺.53ⴱ ⫺1.08ⴱ
Note. X ⫽ independent variable; LMX ⫽ leader-member exchange quality.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.
 14 HUANG, XU, HUANG, AND LIU

A 7

6
Voice constructiveness

5
High LMX

4 Low LMX

3
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
This document is copyrighted by the American Psychological Association or one of its allied publishers.

1
Prohibitive voice
Low High

B 7

6
Overall performance

4 High LMX

Low LMX
3

1
Prohibitive voice
Low High

Figure 3. Results of the quadratic-by-linear interactive effect of prohibitive voice and leader-member exchange
quality (LMX) on voice constructiveness and overall performance (Study 3).

of Table 9, the coefficient for voice constructiveness was signifi-
cant for overall performance (␤ ⫽ .16, p ⬍ .05), while the
quadratic-by-linear interaction of prohibitive voice and LMX be-
came nonsignificant. These findings provided initial support for
the meditating role of voice constructiveness for prohibitive voice
(Hypothesis 4b).
To furthertest Hypothesis 4b, we performed Selig and
Preacher’s (2008) Monte Carlo analysis to examine the moder-
ated mediation model for prohibitive voice. As shown in Table
10, when LMX was low, the very low (␤ ⫽ .22, p ⬍ .05) and
low (␤ ⫽ .10, p ⬍ .05) frequencies of prohibitive voice were
positively and indirectly related to manager-rated overall
performance through voice constructiveness; the medium fre-
quency of prohibitive voice (␤ ⫽ ⫺.02, ns) was not signifi-
cantly related to performance; and the high (␤ ⫽ ⫺.14, p ⬍ .05)
and very high (␤ ⫽ ⫺.26, p ⬍ .05) frequencies of prohibitive
voice were negatively and indirectly related to performance
through voice constructiveness. In contrast, when LMX was
high, voice constructiveness did not mediate the effect of pro-
hibitive voice on overall performance. These findings supported
Hypothesis 4b. Our finding that LMX moderated the mediating
effect of voice constructiveness only for prohibitive voice and
not for promotive voice provided support for Hypothesis 4c.
 FREQUENCY OF VOICE AND MANAGERS’ RESPONSES 15

Table 9
Hierarchical Multilevel Analyses for Overall Performance (Study 3)

Overall performance
Variables M1 M2 M3 M4 M5 M6 M7

Employee age .00 (.02) .00 (.02) ⫺.00 (.02) ⫺.01 (.02) ⫺.01 (.02) ⫺.01 (.02) ⫺.01 (.02)
Employee gender ⫺.03 (.19) ⫺.02 (.19) ⫺.01 (.19) ⫺.05 (.19) ⫺.06 (.19) ⫺.09 (.18) ⫺.09 (.18)
Employee education .02 (.17) .03 (.17) ⫺.02 (.17) .00 (.17) ⫺.03 (.17) ⫺.03 (.16) ⫺.03 (.16)
Employee tenure ⫺.02 (.05) ⫺.03 (.05) ⫺.01 (.05) ⫺.01 (.04) ⫺.00 (.05) ⫺.00 (.04) ⫺.00 (.04)
Dyadic tenure .02 (.08) .02 (.08) .00 (.08) .02 (.08) .00 (.08) .01 (.08) ⫺.00 (.07)
Past performance .07 (.07) .04 (.07) .05 (.07) .04 (.07) ⫺.00 (.07) ⫺.00 (.07) ⫺.03 (.07)
Proactivity ⫺.23ⴱ (.09) ⫺.27ⴱⴱ (.09) ⫺.27ⴱⴱ (.09) ⫺.24ⴱⴱ (.09) ⫺.26ⴱⴱ (.09) ⫺.26ⴱⴱ (.09) ⫺.27ⴱⴱ (.09)
Negative affectivity ⫺.06 (.14) ⫺.08 (.13) ⫺.09 (.14) ⫺.09 (.14) ⫺.12 (.13) ⫺.11 (.13) ⫺.11 (.13)
Managerial openness ⫺.06 (.26) ⫺.06 (.26) ⫺.07 (.26) ⫺.07 (.27) ⫺.04 (.26) ⫺.04 (.25) .01 (.25)
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Promotive voice (PMV) .14 (.10) .12 (.10) .11 (.10) .07 (.10) .09 (.10) .09 (.10)
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Prohibitive voice (PHV) .12 (.11) .15 (.11) .15 (.11) .15 (.10) .14 (.10) .14 (.10)
PMV2 ⫺.08 (.07) ⫺.10 (.07) ⫺.09 (.07) ⫺.07 (.07) ⫺.07 (.07)
PHV2 ⫺.13 (.09) ⫺.11 (.09) ⫺.09 (.09) ⫺.07 (.09) ⫺.06 (.09)
LMX .21ⴱ (.10) .20ⴱ (.09) ⫺.02 (.15) ⫺.02 (.15)
PMV ⫻ LMX .21ⴱ (.10) .24ⴱ (.10) .22ⴱ (.10)
PHV ⫻ LMX .15 (.09) .13 (.09) .10 (.09)
PMV2 ⫻ LMX ⫺.02 (.07) ⫺.05 (.07)
PHV2 ⫻ LMX .20ⴱ (.09) .15 (.09)
Mediator
Voice constructiveness .16ⴱ (.07)
⌬␹2 (df) 7.19 (9) 4.44 (2) 5.22 (2) 5.02ⴱ (1) 6.20ⴱ (2) 5.44 (2) 4.41ⴱ (1)
Pseudo ⌬R2 .04 .03 .01 .02 .04 .04 .03
Note. M ⫽ Model; LMX ⫽ leader-member exchange quality.
ⴱ
p ⬍ .05. ⴱⴱ p ⬍ .01. ⴱⴱⴱ p ⬍ .001.

Discussion hibitive voice on manager-rated overall performance of voicers.

In Study 3, we not only replicated the findings of Studies 1 and
2 that supported Hypotheses 2b and 2c, but also obtained empirical
evidence supporting Hypotheses 3b, 3c, 4b, and 4c. These findings
support our key proposition that the frequency of challenging
voice, voice content, and LMX context jointly predict managers’
evaluations of voicing employees through influencing how man-
agers perceive the constructiveness of voice. In Study 3, the
pattern of the interactive effects of prohibitive voice and LMX on
manager-perceived voice constructiveness and manager-rated
overall performance was very similar to the pattern of those effects
on manager-rated promotability found in Studies 1 and 2. This
result remained robust even when we included a few more control
variables, such as employee proactivity, managers’ openness, and
managers’ negative affectivity.
Overall Discussion
The central contention of the current research is that the
frequency of challenging voice and managers’ evaluations of
voicing employees are not related in a linear manner, as previ-
ously assumed (Burris, 2012). Rather, as findings across our
three studies have largely supported, there exists an inverted
U-shaped relationship between the frequency of challenging
voice and managers’ ratings of voicers’ promotability (Studies
1 and 2) and their overall performance evaluations (Study 3),
especially when the content of challenging voice is prohibitive
rather than promotive in nature and when the voicers have a low
rather than high level of LMX with managers. In Study 3, we
also found that, in the case of low LMX, manager-perceived
voice constructiveness mediated the curvilinear effect of pro-
A summary of our findings is shown in Table 11. In general, six
out of the 12 hypotheses (Hypotheses 2b, 2c, 3b, 3c, 4b, and 4c)
were supported across three studies, and two hypotheses (Hy-
potheses 1b and 1c) received support only in Study 2.3
The findings of our study provide initial evidence for the cur-
vilinear effect of challenging voice, and therefore, they have
several important implications for research on employee voice.
First, drawing from social persuasion theory (e.g., Briñol & Petty,
2009; Eagly & Chaiken, 1998), we extended Whiting et al.’s
(2012) experimental work on managers’ responses to voice by
examining the joint effects of the frequency of voice (source
characteristic), content of challenging voice (message characteris-
tic), and relational context of LMX (context characteristic) that are
relevant to organizational settings. Frequency of challenging voice
provides signals to managers about voicing employees’ degree of
effort to initiate changes for the organization functioning. Promo-
tive voice and prohibitive voice capture major contents of chal-
lenging voice (Liang et al., 2012). LMX serves as a salient bound-
ary condition that reflects the relational context for voice behavior.
Results across the three independent samples consistently showed
that these three variables jointly predict the extent to which man-
3
To examine whether the samples of three studies had sufficient statis-
tical power to replicate the complex model, we followed the procedure
suggested by Faul, Erdfelder, Lang, and Buchner (2007) to perform a set
of post hoc power analyses in all three samples. Specifically, we tested the
statistical power for the moderating effects of LMX on the curvilinear link
between promotive/prohibitive voice frequency and managers’ evalua-
tions. Results showed a sufficient statistical power for all three studies
(Study 1 ⫽ .99; Study 2 ⫽ .99, Study 3 ⫽ .82, all above the threshold of
.80).
 16 HUANG, XU, HUANG, AND LIU

Table 10 koff, 2014; Whiting et al., 2012) and recipients’ characteristics

Conditional Indirect Effect of Prohibitive Voice on Manger- (Fast, Burris, & Bartel, 2014; Whiting et al., 2012). Departing
Rated Overall Performance Through Manager-Perceived Voice from this linear approach to understanding the consequences of
Constructiveness (Study 3)a voice, our studies have demonstrated the critical yet overlooked
role of the frequency of challenging voice and its nonlinear effect
Moderator Predictor Overall performance on managers’ evaluations of voicing employees. To develop more
LMX Voice Indirect effect 95% confidence interval precise predictions of the consequences of challenging voice,
future research should model this nonlinear effect of voice fre-
High Very low b
⫺.14 ⫺.39, .03
High Low ⫺.06 ⫺.20, .03 quency.
High Medium .01 ⫺.03, .07 Third, consistent with social persuasion theory (Briñol & Petty,
High High .09 ⫺.00, .23 2009) and findings of previous studies on the consequences of
High Very high .16 ⫺.00, .42 voice (Burris, 2012; Chamberlin et al., 2017; Maynes & Podsa-
Low Very low .22ⴱ .02, .49
Low Low .10ⴱ .01, .24 koff, 2014), we found that the content of challenging voice plays
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Low Medium ⫺.02 ⫺.07, .02 an important role in determining managers’ evaluations. In addi-
This document is copyrighted by the American Psychological Association or one of its allied publishers.

Low High ⫺.14ⴱ ⫺.31, ⫺.02
Low Very high ⫺.26ⴱ ⫺.56, ⫺.03
Note. LMX ⫽ leader-member exchange quality.
a
Based on 20,000 Monte Carlo samples (Selig & Preacher, 2008). b Very
high ⫽ 2 SD above the mean; High ⫽ 1 SD above the mean; Medium ⫽
mean value; Low ⫽ 1 SD below the mean; Very low ⫽ 2 SD below the
mean.
ⴱ
p ⬍ .05.
agers reward voicing employees by giving them positive evalua-
tions.
Second, our findings suggest that the mixed results regarding the
consequences of challenging voice reported in the literature can be
explained not only by identifying its boundary conditions but also
by modeling a nonlinear effect of the frequency of voice. Specif-
ically, some previous studies have reported that challenging voice
sometimes receives positive reactions (Burris, 2012; Whiting et al.,
2012), whereas other studies have reported that it receives negative
reactions (Burris, 2012; Seibert et al., 2001) or even no reaction
(Van Dyne & LePine, 1998) from managers. Assuming a linear
effect of challenging voice on managers’ evaluations of voicing
employees, several studies have advanced our understanding of the
boundary conditions of the consequences of challenging voice by
identifying certain moderators, such as different features of voice
content (Burris, 2012; Chamberlin et al., 2017; Maynes & Podsa-
tion to taking the content of challenging voice (promotive and
prohibitive) into account as Chamberlin et al. (2017) suggested,
our results show that it is also important to model the differential
effects of the two types of voice in a nonlinear manner. Indeed,
across the three studies, we found a significant quadratic-by-linear
interactive effect of prohibitive voice and LMX on managers’
evaluations, although we did not find a nonlinear effect of promo-
tive voice in either high or low LMX dyads. In other words, the
hypothesized inverted U-shaped effect can be observed for pro-
hibitive voice but not for promotive voice in low LMX dyads.
Although Chamberlin et al.’s (2017) meta-analytical study clearly
shows that prohibitive voice tends to lead to managers’ negative
evaluations in general, our nonlinear model has further advanced
our understanding of managers’ responses to prohibitive voice by
showing that managers may appreciate prohibitive voice from
outgroup members when these employees refrain from speaking
out very frequently.
It is theoretically interesting to contemplate why we did not find
consistently significant effects of promotive voice. Because pro-
motive voice focuses on suggesting solutions and new ideas,
perhaps the quality of these solutions and ideas may be more
critical in affecting managers’ evaluations of their constructiveness
as well as their evaluations of the voicing employees. It is easier to
point out problems than to come up with good solutions. Thus, it
is possible that promotive voice induces managers’ positive eval-

Table 11
Summary of the Results of Hypothesis Testing

Study 1 Study 2 Study 3

(Chinese (U.S. (U.S.
sample) sample) sample)
Hypotheses results results results

H1a (curvilinear effect of promotive voice) NSa NS NS

H1b (curvilinear effect of prohibitive voice) NS S NS
H1c (comparative effects of two types of voice) NS S NS
H2a (curvilinear interactive effect of promotive voice and LMX on managers’ evaluations) NS NS NS
H2b (curvilinear interactive effect of LMX and prohibitive voice and LMX on managers’ evaluations) S S S
H2c (comparative moderating effects of LMX on two types of voice for managers’ evaluations) S S S
H3a (curvilinear interactive effect of promotive voice and LMX on voice constructiveness) NS
H3b (curvilinear interactive effect of prohibitive voice and LMX on voice constructiveness) S
H3c (comparative moderating effects of LMX on two types of voice for voice constructiveness) S
H4a (Mediating Role of Voice Constructiveness for Promotive Voice ⫻ LMX) NS
H4b (Mediating Role of Voice Constructiveness for Prohibitive Voice ⫻ LMX) S
H4c (comparative mediating role of voice constructiveness for both types of voice in high vs low LMX dyads) S
Note. S ⫽ supported; NS ⫽ not supported; LMX ⫽ leader-member exchange quality.
 FREQUENCY OF VOICE AND MANAGERS’ RESPONSES
uation primarily for those voicers who often come up with high-
quality suggestions. Indeed, in Study 3, we found a significant
linear interaction of promotive voice and LMX, in that the positive
effect of promotive voice frequency on manager-rated overall
performance is stronger in high LMX dyads than in low ones. This
is perhaps because managers tend to make positive assessments of
the voice quality of ingroup members. As LMX may not fully
capture the quality of voice from ingroup employees, we did not
find this linear interaction consistently across the three samples. To
confirm this speculation, future research should directly examine
whether the effect of promotive voice is shaped by moderators,
such as perceived quality of voice.
Fourth, our findings highlight the role of LMX in shaping the
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
consequences of challenging voice for managers’ evaluations of
This document is copyrighted by the American Psychological Association or one of its allied publishers.
voicers. Social persuasion theory suggests that people are more
likely to accept influence from their ingroup members than from
outgroup members (e.g., Briñol & Petty, 2009; Clark & Maass,
1988; Crano & Chen, 1998; Menon & Blount, 2003). Across all
studies, we found that LMX was strongly and positively correlated
with managers’ evaluations of employees and perceptions of voice
constructiveness. Moreover, the moderating effects of LMX found
in the three studies point to the overriding role of LMX in shaping
managers’ positive responses to challenging voice. It therefore
would be desirable for future research to take into account the
influence of LMX when investigating the consequences of chal-
lenging voice for managers’ evaluations.
Fifth, Study 3 demonstrated that manager-perceived voice con-
structiveness is a key mechanism that may help explain the joint
effects of voice frequency, voice content, and LMX on managers’
evaluations. Also, as shown in Table 10, LMX had a strong main
effect on voice constructiveness, corroborating our key logic:
Relational contexts shape managers’ cognitive assessments of the
beneficial outcomes of voice, which in turn influences managers’
responses. These findings extend the cognition in social persuasion
model to organizational settings (Albarracín, 2002; Albarracín &
Wyer, 2001; Hart et al., 2009), by suggesting that managers’
cognitive evaluations of the benefits of voiced issues are a salient
mechanism explaining the consequences of challenging voice in
organizations.
Finally, in Studies 2 and 3, we tested our proposed model using
U.S. samples and found similar results as those reported in Study
1 that used a Chinese sample. The findings of the three studies
offer evidence for the cross-cultural generalizability of our model.
Given the large power distance culture of China, subordination is
socially expected by both managers and employees (House,
Hanges, Javidan, Dorfman, & Gupta, 2004). One could argue that,
in such a context, Chinese managers may be less accepting of
challenging voice from employees than are their counterparts in
the West. Likewise, managers from a small-power-distance culture
may not see engaging in challenging voice as an act of insubor-
dination; instead, they may regard such behavior as constructive to
organizations. Countering these views, we indeed found a signif-
icant main curvilinear effect of prohibitive voice in Study 2 and a
significant quadratic-by-linear interactive effect of prohibitive
voice and LMX in Studies 2 and 3. These findings suggest that
U.S. managers, like their Chinese counterparts, do not tolerate high
frequency prohibitive voice from outgroup employees. Although
we are not entirely certain that our findings can be generalized to
nations other than China and the United States, based on our
17
theoretical reasoning and the robust empirical findings, we do
expect that this curvilinear effect of challenging voice would be
found in different cultures in future research.
Limitations and Future Research
Despite the strengths of using three independent samples, time-
lagged surveys in all studies, and both supervisor-rated (Studies 1
and 3) and subordinate-rated (Study 2) voice, our research has
three limitations. First, in Studies 1 and 3, our independent vari-
ables, the mediator, and the dependent variables came from the
same source: the supervisors. This may raise concerns about com-
mon method bias. To partially address this limitation, we used
subordinate-reported measures of challenging voice in Study 2 and
found results similar to those in Studies 1 and 3. Using
subordinate-rated voice remains problematic too, however. It is
methodologically reasonable to ask supervisors to report their
perceived employees’ voice behavior. In reality, managers some-
times may not be aware of employees’ voice behavior. More
important, according to our theorization, managers’ evaluations of
voicing employees are driven by their observations of particular
employees’ frequency of voice. To establish this relationship, we
need to make sure that managers are aware of their employees’
voice behavior. If we ask employees to report their own voice
behavior, we cannot fully capture managers’ perceived frequency.
To accurately measure employee voice while also avoiding com-
mon source problems, future research should consider using lab-
based studies to manipulate employee voice frequency.
Second, managers’ evaluations of promotability, overall perfor-
mance, and voice constructiveness could be influenced by another
key voice characteristic: the quality of voice. Quality of messages,
which is often operationalized in terms of the strength of argu-
ments, has been widely used in experimental settings to predict
message receivers’ reactions to a single “incident” of persuasion
(Briñol & Petty, 2009; Eagly & Chaiken, 1998). In organizational
settings, however, managers’ evaluations of voicing employees are
less likely to be influenced by the quality of a single incident of
voice. Also, it is difficult to judge the quality of voiced messages,
because, unlike in experimental settings, it may take time for
managers to learn the quality of voice from employees. In field
settings, voice quality can be conceptualized as either a voicer
characteristic or a voice characteristic. It is a voicer characteristic
when it is conceptualized as an employee’s “history” of voice
quality. It becomes a voice characteristic if the quality of voice is
conceptualized in terms of whether the voiced issues are within the
voicers’ functional area of expertise (Dutton & Ashford, 1993).
Given its conceptual and operationalization complexity in field
settings, we did not incorporate quality of voice in our model.
Nevertheless, we encourage future researchers to theorize and
investigate how quality of voice plays a role in the nonlinear
relationship between frequency of challenging voice and manag-
ers’ evaluations of voicers.
Finally, voice constructiveness was measured using a two-item
scale adopted from Gorden (1988). Although other researchers
also have used this scale to capture cognitive processes underlying
the effects of voice (e.g., Maynes & Podsakoff, 2014; Whiting et
al., 2012), the two-item measure may not fully capture the cogni-
tive assessments of the beneficial outcomes of voice for organiza-
 18 HUANG, XU, HUANG, AND LIU
tions. Future research should develop a scale that can better reflect
this cognitive process.
Our results may suggest some new directions for studying the
effects of challenging voice on managers’ evaluations of voicers.
First, we cast LMX as a key relational context moderator in our
model, focusing on how managers respond differentially to in-
group voicers and outgroup voicers. Research studies, however,
have suggested that the triadic relationship between a manager, an
ingroup subordinate, and an outgroup subordinate can be more
complex than previous research on LMX has assumed (Sherony &
Green, 2002; Tse, Lam, Lawrence, & Huang, 2013). The manag-
er’s responses to the voice behavior of an ingroup subordinate may
be influenced by the outgroup subordinate’s responses to the voice
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
behavior, or vice versa. Future research could extend our model by
This document is copyrighted by the American Psychological Association or one of its allied publishers.
examining the potential influences from a third party in the work
group. Second, in addition to message, source, and context factors,
social persuasion theory also suggests that receiver characteristics
play a key role in receivers’ reactions to persuasion (McGuire,
1985). Thus, another direction of expanding our model would be to
consider a set of managers’ key characteristics (receiver factors)
such as managers’ proactive personality (Parker, 1998), that may
influence the curvilinear effects of challenging voice.
Practical Implications
Our findings have several useful implications for practitioners.
First, it is essential for subordinates to understand the true risks of
voicing out challenging issues to their managers. The findings of
the current study suggest that challenging voice, such as prohibi-
tive voice, may not be as risky as previously assumed. A moderate
frequency of prohibitive voice can project a positive image to
managers, even when the voicing employees are not members of
the managers’ ingroup. More important, subordinates who have
low LMX with their managers should be aware that only very
frequent prohibitive voice tends to induce managers’ unfavorable
responses. Second, in concert with what we know from the LMX
and voice literatures, the most effective way for subordinates to
influence managers is to develop high levels of social exchange
relationships with them. Third, managers can learn from our study
that the frequency of employees’ voicing may influence their
evaluations of the voiced messages, causing them to overlook
some very important issues. Managers tend to have unfavorable
evaluations of the voice of outgroup members in general, yet our
findings show that managers may indeed appreciate voice from
outgroup members when these members voice up challenging
issues at a moderate frequency. Managers, however, tend to de-
grade those outgroup members who voice up challenging issues
very frequently. As such, managers may inadvertently discourage
outgroup members from raising important issues and miss oppor-
tunities to identify critical problems that may affect their organi-
zations’ functioning.
References
Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and
interpreting interactions. Thousand Oaks, CA: Sage.
Albarracín, D. (2002). Cognition in persuasion: An analysis of information
processing in response to persuasive communications. In M. P. Zanna
(Ed.), Advances in experimental social psychology (Vol. 34, pp. 61–
130). San Diego, CA: Academic Press.
Albarracín, D., & Wyer, R. S., Jr. (2001). Elaborative and nonelaborative
processing of a behavior-related communication. Personality and Social
Psychology Bulletin, 27, 691–705. http://dx.doi.org/10.1177/
0146167201276005
Bartko, J. J. (1976). On various intraclass correlation reliability coeffi-
cients. Psychological Bulletin, 83, 762–765. http://dx.doi.org/10.1037/
0033-2909.83.5.762
Bohner, G., & Dickel, N. (2011). Attitudes and attitude change. Annual
Review of Psychology, 62, 391– 417. http://dx.doi.org/10.1146/annurev
.psych.121208.131609
Briñol, P., & Petty, R. E. (2009). Source factors in persuasion: A self-
validation approach. European Review of Social Psychology, 20, 49 –96.
http://dx.doi.org/10.1080/10463280802643640
Brislin, R., Lonner, W. J., & Thorndike, R. (1973). Cross-cultural research
methods. New York, NY: Wiley.
Burris, E. R. (2012). The risks and rewards of speaking up: Managerial
responses to employee voice. Academy of Management Journal, 55,
851– 875. http://dx.doi.org/10.5465/amj.2010.0562
Burris, R. B., Detert, J. R., & Romney, A. C. (2013). Speaking up vs. being
heard: The disagreement around and outcomes of employee voice.
Organization Science, 24, 22–38. http://dx.doi.org/10.1287/orsc.1110
.0732
Chamberlin, M., Newton, D. W., & Lepine, J. A. (2017). A meta-analysis
of voice and its promotive and prohibitive forms: Identification of key
associations, distinctions, and future research directions. Personnel Psy-
chology, 70, 11–71. http://dx.doi.org/10.1111/peps.12185
Clark, R. D., & Maass, A. (1988). Social categorization in minority
influence: The case of homosexuality. European Journal of Social
Psychology, 18, 347–364. http://dx.doi.org/10.1002/ejsp.2420180405
Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple
regression/correlation analysis for the behavioral sciences (3rd ed.).
London, UK: Erlbaum.
Crano, W. D., & Chen, X. (1998). The leniency contract and persistence of
majority and minority influence. Journal of Personality and Social
Psychology, 74, 1437–1450. http://dx.doi.org/10.1037/0022-3514.74.6
.1437
De Stobbeleir, K. E. M., Ashford, S. J., & de Luque, M. F. S. (2010).
Proactivity with image in mind: How employee and manager character-
istics affect evaluations of proactive behaviours. Journal of Occupa-
tional and Organizational Psychology, 83, 347–369. http://dx.doi.org/
10.1348/096317909X479529
Detert, J. R., & Burris, E. R. (2007). Leadership behavior and employee
voice: Is the door really open? Academy of Management Journal, 50,
869 – 884. http://dx.doi.org/10.5465/AMJ.2007.26279183
Dutton, J. E., & Ashford, S. J. (1993). Selling issues to top management.
The Academy of Management Review, 18, 397– 428.
Eagly, A. H., & Chaiken, S. (1998). Attitude structure and function. In
D. T. Gilbert, S. T. Fiske, & G. Lindzey (Eds.), The handbook of social
psychology (Vol. 1–2, pp.269 –322). New York, NY: McGraw-Hill.
Eagly, A., Wood, W., & Chaiken, S. (1978). Causal inferences about
communicators and their effect on opinion change. Journal of Person-
ality and Social Psychology, 36, 424 – 435. http://dx.doi.org/10.1037/
0022-3514.36.4.424
Fast, N. J., Burris, E. R., & Bartel, C. A. (2014). Managing to stay in the
dark: Managerial self-efficacy, ego defensiveness, and the aversion to
employee voice. Academy of Management Journal, 57, 1013–1034.
http://dx.doi.org/10.5465/amj.2012.0393
Faul, F., Erdfelder, E., Lang, A. G., & Buchner, A. (2007). GⴱPower 3: A
flexible statistical power analysis program for the social, behavioral, and
biomedical sciences. Behavior Research Methods, 39, 175–191. http://
dx.doi.org/10.3758/BF03193146
Fuller, B., Marler, L. E., Hester, K., & Otondo, R. F. (2015). Leader
reactions to follower proactive behavior: Giving credit when credit is
 FREQUENCY OF VOICE AND MANAGERS’ RESPONSES
due. Human Relations, 68, 879 – 898. http://dx.doi.org/10.1177/
0018726714548235
Gorden, W. I. (1988). Range of employee voice. Employee Responsibilities
and Rights Journal, 1, 283–299. http://dx.doi.org/10.1007/BF01556937
Gosling, S. D., Rentfrow, P. J., & Swann, W. B., Jr. (2003). A very brief measure
of the Big-Five personality domains. Journal of Research in Personality, 37,
504–528. http://dx.doi.org/10.1016/S0092-6566(03)00046-1
Graen, G. B., & Uhl-Bien, M. (1995). Relationship-based approach to
leadership: Development of leader-member exchange (LMX) theory of
leadership over 25 years: Applying a multi-level multi-domain perspec-
tive. The Leadership Quarterly, 6, 219 –247. http://dx.doi.org/10.1016/
1048-9843(95)90036-5
Grant, A. M., Parker, S., & Collins, C. (2009). Getting credit for proactive
behavior: Supervisor reactions depend on what you value and how you
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
feel. Personnel Psychology, 62, 31–55. http://dx.doi.org/10.1111/j.1744-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
6570.2008.01128.x
Hart, W., Albarracín, D., Eagly, A. H., Brechan, I., Lindberg, M. J., &
Merrill, L. (2009). Feeling validated versus being correct: A meta-
analysis of selective exposure to information. Psychological Bulletin,
135, 555–588. http://dx.doi.org/10.1037/a0015701
Hofstede, G. (2001). Culture’s consequences: Comparing values, behav-
iors, institutions, and organizations across nations. Thousand Oaks, CA:
Sage.
House, R. J., Hanges, P. J., Javidan, M., Dorfman, P. W., & Gupta, V.
(2004). Culture, leadership, and organizations: The globe study of 62
societies. Thousand Oaks, CA: Sage.
Hovland, C. I., & Mandell, W. (1952). An experimental comparison of
conclusion-drawing by the communicator and by the audience. The
Journal of Abnormal and Social Psychology, 47, 581–588. http://dx.doi
.org/10.1037/h0059833
Howard-Grenville, J. A. (2007). Developing issue-selling effectiveness
over time: Issue selling as resourcing. Organization Science, 18, 560 –
577. http://dx.doi.org/10.1287/orsc.1070.0266
Huang, X., Van de Vliert, E., & Van der Vegt, G. (2005). Breaking the
culture of silence: Does the stimulation of participation reduce employee
opinion withholding cross-nationally? Management and Organization
Review, 1, 459 – 482. http://dx.doi.org/10.1111/j.1740-8784.2005
.00023.x
Isenberg, D. J. (1986). Group polarization: A critical review and meta-
analysis. Journal of Personality and Social Psychology, 50, 1141–1151.
http://dx.doi.org/10.1037/0022-3514.50.6.1141
Johnson, J. W. (2000). A heuristic method for estimating the relative weight of
predictor variables in multiple regression. Multivariate Behavioral Research,
35, 1–19. http://dx.doi.org/10.1207/S15327906MBR3501_1
Jones, L. W., Sinclair, R. C., & Courneya, K. S. (2003). The effects of
source credibility and message framing on exercise intentions, behav-
iors, and attitudes: An integration of the elaboration likelihood model
and prospect theory. Journal of Applied Social Psychology, 33, 179 –
196. http://dx.doi.org/10.1111/j.1559-1816.2003.tb02078.x
Lam, C. F., Rees, L., Levesque, L. L., & Ornstein, S. (in press). Shooting
from the hip: A habit perspective of voice. Academy of Management
Review.
Lam, W., Huang, X., & Snape, E. (2007). Feedback-seeking behavior and
leader-member exchange: Does supervisor-attributed motive matter?
Academy of Management Journal, 50, 348 –363. http://dx.doi.org/10
.5465/AMJ.2007.24634440
Levin, I. P., Schneider, S. L., & Gaeth, G. J. (1998). All frames are not
created equal: A typology and critical analysis of framing effects.
Organizational Behavior and Human Decision Processes, 76, 149 –188.
http://dx.doi.org/10.1006/obhd.1998.2804
Li, A. N., Liao, H., Tangirala, S., & Firth, B. M. (2017). The content of the
message matters: The differential effects of promotive and prohibitive
team voice on team productivity and safety performance gains. Journal
19
of Applied Psychology, 102, 1259 –1270. http://dx.doi.org/10.1037/
apl0000215
Li, N., Liang, J., & Crant, J. M. (2010). The role of proactive personality
in job satisfaction and organizational citizenship behavior: A relational
perspective. Journal of Applied Psychology, 95, 395– 404. http://dx.doi
.org/10.1037/a0018079
Liang, J., Farh, C. I. C., & Farh, J. L. (2012). Psychological antecedents of
promotive and prohibitive voice: A two-wave examination. Academy of
Management Journal, 55, 71–92. http://dx.doi.org/10.5465/amj.2010
.0176
Liden, R. C., Sparrowe, R. T., & Wayne, S. J. (1997). Leader-member
exchange theory: The past and potential for the future. In G. R. Ferris
(Ed.), Research in personnel and human resources management (Vol.
15, pp. 47–119). Greenwich, CT: JAI.
Liden, R. C., Wayne, S. J., & Stilwell, D. (1993). A longitudinal study on
the early development of leader-member exchanges. Journal of Applied
Psychology, 78, 662– 674. http://dx.doi.org/10.1037/0021-9010.78.4
.662
Liu, W., Tangirala, S., Lam, W., Chen, Z., Jia, R. T., & Huang, X. (2015).
How and when peers’ positive mood influences employees’ voice be-
havior. Journal of Applied Psychology, 100, 976 –989. http://dx.doi.org/
10.1037/a0038066
MacKenzie, S. B., Podsakoff, P. M., & Fetter, R. (1991). Organizational
citizenship behavior and objective productivity as determinants of man-
agerial evaluations of salespersons’ performance. Organizational Behav-
ior and Human Decision Processes, 50, 123–150. http://dx.doi.org/10
.1016/0749-5978(91)90037-T
Mason, W. A., Conrey, F. R., & Smith, E. R. (2007). Situating social
influence processes: Dynamic, multidirectional flows of influence
within social networks. Personality and Social Psychology Review, 11,
279 –300. http://dx.doi.org/10.1177/1088868307301032
Maynes, T. D., & Podsakoff, P. M. (2014). Speaking more broadly: An
examination of the nature, antecedents, and consequences of an ex-
panded set of employee voice behaviors. Journal of Applied Psychology,
99, 87–112. http://dx.doi.org/10.1037/a0034284
McCroskey, J. C. (1969). A summary of experimental research on the
effect of evidence in persuasive communication. The Quarterly Journal
of Speech, 55, 169 –176. http://dx.doi.org/10.1080/00335636909382942
McGuire, W. J. (1985). Attitudes and attitude change. In G. Lindzey & E.
Aronson (Eds.), Handbook of social psychology (pp. 233–346). New
York, NY: Random House.
Menon, T., & Blount, S. (2003). The messenger bias: A relational model of
knowledge valuation. Research in Organizational Behavior, 25, 137–
186. http://dx.doi.org/10.1016/S0191-3085(03)25004-8
Milliken, F. J., Morrison, E. W., & Hewlin, P. F. (2003). An exploratory
study of employee silence: Issues that employees don’t communicate
upward and why. Journal of Management Studies, 40, 1453–1476.
http://dx.doi.org/10.1111/1467-6486.00387
Morrison, E. W. (2014). Employee voice and silence. Annual Review of
Organizational Psychology and Organizational Behavior, 1, 173–197.
http://dx.doi.org/10.1146/annurev-orgpsych-031413-091328
Morrison, E. W., & Milliken, F. J. (2003). Speaking up, remaining silent: The
dynamics of voice and silence in organizations. Journal of Management
Studies, 40, 1353–1358. http://dx.doi.org/10.1111/1467-6486.00383
Motowidlo, S. J., & Van Scotter, J. R. (1994). Evidence that task performance
should be distinguished from contextual performance. Journal of Applied
Psychology, 79, 475– 480. http://dx.doi.org/10.1037/0021-9010.79.4.475
Parker, S. K. (1998). Enhancing role breadth self-efficacy: The roles of job
enrichment and other organizational interventions. Journal of Applied Psy-
chology, 83, 835– 852. http://dx.doi.org/10.1037/0021-9010.83.6.835
Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of
persuasion. In L. Berkowitz (Ed.), Advances in experimental social psy-
chology (Vol. 19, pp. 123–205). New York, NY: Academic Press.
 20 HUANG, XU, HUANG, AND LIU
Petty, R. E., Cacioppo, J. T., & Goldman, R. (1981). Personal involvement
as a determinant of argument-based persuasion. Journal of Personality
and Social Psychology, 41, 847– 855. http://dx.doi.org/10.1037/0022-
3514.41.5.847
Raudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models:
Applications and data analysis methods (2nd ed.). Thousand Oaks, CA:
Sage.
Scandura, T. A., & Schriesheim, C. A. (1994). Leader-member exchange
and supervisor career mentoring. Academy of Management Journal, 37,
1588 –1602. http://dx.doi.org/10.2307/256800
Schilit, W. K., & Locke, E. A. (1982). A study of upward influence in
organizations. Administrative Science Quarterly, 27, 304 –316. http://dx
.doi.org/10.2307/2392305
Seibert, S., Kraimer, M., & Crant, J. (2001). What do proactive people do?
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
A longitudinal model linking proactive personality and career success.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
Personnel Psychology, 54, 845– 874. http://dx.doi.org/10.1111/j.1744-
6570.2001.tb00234.x
Selig, J. P., & Preacher, K. J. (2008). Monte Carlo method for assessing
mediation: An interactive tool for creating confidence intervals for
indirect effects [Computer software]. Retrieved from from http://
quantpsy.org/
Sherony, K. M., & Green, S. G. (2002). Coworker exchange: Relationships
between coworkers, leader-member exchange, and work attitudes. Jour-
nal of Applied Psychology, 87, 542–548. http://dx.doi.org/10.1037/0021-
9010.87.3.542
Tangirala, S., & Ramanujam, R. (2012). Ask and you shall hear (But not
always): Examining the relationship between manager consultation and
employee voice. Personnel Psychology, 65, 251–282. http://dx.doi.org/
10.1111/j.1744-6570.2012.01248.x
Thomas, J. P., Whitman, D. S., & Viswesvaran, C. (2010). Employee
proactivity in organizations: A comparative meta-analysis of emergent
proactive constructs. Journal of Occupational and Organizational Psy-
chology, 83, 275–300. http://dx.doi.org/10.1348/096317910X502359
Thompson, E. R. (2007). Development and validation of an internationally
reliable short-form of the Positive and Negative Affect Schedule (PA-
NAS). Journal of Cross-Cultural Psychology, 38, 227–242. http://dx.doi
.org/10.1177/0022022106297301
Tonidandel, S., & LeBreton, J. M. (2015). RWA web: A free, comprehen-
sive, web-based, and user-friendly tool for relative weight analyses.
View publication stats
Journal of Business and Psychology, 30, 207–216. http://dx.doi.org/10
.1007/s10869-014-9351-z
Tost, L. P., Gino, F., & Larrick, R. P. (2012). Power, competitiveness, and
advice taking: Why the powerful don’t listen. Organizational Behavior
and Human Decision Processes, 117, 53– 65. http://dx.doi.org/10.1016/
j.obhdp.2011.10.001
Tse, H. H., Lam, C. K., Lawrence, S. A., & Huang, X. (2013). When my
supervisor dislikes you more than me: The effect of dissimilarity in
leader-member exchange on coworkers’ interpersonal emotion and per-
ceived help. Journal of Applied Psychology, 98, 974 –988. http://dx.doi
.org/10.1037/a0033862
Uhl-Bien, M., & Maslyn, J. M. (2003). Reciprocity in manager-
subordinate relationships: Components, configurations, and out-
comes. Journal of Management, 29, 511–532. http://dx.doi.org/10
.1016/S0149-2063_03_00023-0
Van Dyne, L., Cummings, L. L., & McLean Parks, J. (1995). Extra-role
behaviors: In pursuit of construct and definitional clarity. In L. L.
Cummings & B. M. Staw (Eds.), Research in organizational behavior
(Vol. 17, pp. 215–285). Greenwich, CT: JAI Press.
Van Dyne, L., & LePine, J. A. (1998). Helping and voice extra-role
behaviors: Evidence of construct and predictive validity. Academy of
Management Journal, 41, 108 –119. http://dx.doi.org/10.2307/256902
Vaniv, I., & Choshen-Hillel, S. (2012). When guessing what another
person would say is better than giving your own opinion: Using
perspective-taking to improve advice-taking. Journal of Experimental
Social Psychology, 48, 1022–1028. http://dx.doi.org/10.1016/j.jesp.2012
.03.016
Whiting, S. W., Maynes, T. D., Podsakoff, N. P., & Podsakoff, P. M.
(2012). Effects of message, source, and context on evaluations of em-
ployee voice behavior. Journal of Applied Psychology, 97, 159 –182.
http://dx.doi.org/10.1037/a0024871
Whiting, S. W., Podsakoff, P. M., & Pierce, J. R. (2008). Effects of task
performance, helping, voice, and organizational loyalty on performance
appraisal ratings. Journal of Applied Psychology, 93, 125–139. http://dx
.doi.org/10.1037/0021-9010.93.1.125
Received January 18, 2016
Revision received April 17, 2018
Accepted April 20, 2018 䡲